(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Professional standards review organization : program evaluation"

Report 



Professional Standards Review 
Organization 1979 Program Evaluation 



Published by the Health Care Financing; AdminiHtration 



Health Care 

Financing 

Research 

Report 



Health Care Financing Research Reports are published on a periodic basis by the Health 
Care Financing Administration's Office of Research, Demonstrations, and Statistics as inter- 
nal projects are completed. 

The Health Care Financing Administration was established in March 1977 to combine HEW's 
health financing and quality assurance programs into a single agency. HCFA is responsible 
for the operation of the Medicare and Medicaid programs, the PSRO program, Federal survey 
and certification efforts, and a variety of health care quality assurance activities. 

The mission of the Health Care Financing Administration is to promote the timely delivery of 
appropriate, quality health care to its beneficiaries— approximately 45 million aged, disabled, 
and poor Americans. HCFA is committed to making beneficiaries aware of the services for 
which they are eligible, promoting the accessibility of those services and ensuring that 
HCFA policies and actions promote efficiency and quality within the total health care 
delivery system. 

HCFA's Office of Research, Demonstrations, and Statistics (ORDS) conducts studies and 
projects that demonstrate and evaluate optional reimbursement, coverage, eligibility, and 
management alternatives to the present Federal programs. ORDS also assesses the impact 
of HCFA programs on health care costs, program expenditures, beneficiary access to ser- 
vices, health care providers, and the health care industry. In addition, ORDS monitors na- 
tional health care expenditures and prices and provides actuarial analyses on the costs of 
current HCFA programs as well as the impact of possible legislative or administrative 
changes in the programs. 

Health Care Financing Research Reports present the results of major studies and projects 
conducted by the HCFA program staff. These reports contain significant findings which af- 
fect HCFA programs and are used as the basis for making program changes. 



RA 



I'^v'fHealth Care 

Financing 

Research 

Report 



Professional Standards Review Organization 
1979 Program Evaluation 



OA5 Ubi-ii'v 

C2"07-13 

7500 Security Blvd. 

Boitimore, MarviSnd 21244 



Published by 

Health Care Financing Administration 
Office of Research, Demonstrations, 
and Statistics 



CONTENTS 



Foreword 

Executive Summary 
Contributors 

1. Introduction and Summary 

1.1 Background 

1.2 Context 

1.2.1 The Problems 

1.2.2 The Response 

1.3 The PSRO Program 

1.3.1 Mandate 

1.3.2 Implementation 

1.3.3 Potential 

1.4 Evaluation Mandate 

1.5 Evaluation Content and Comparison with 
Prior Evaluations 

1.6 Findings 

1.6.1 Medicare Impact Study 

1.6.2 Diagnosis- and Procedure- 
Specific Impact Study 

1.6.3 Medicaid Utilization Rate 
Development Study 

1.6.4 Case-Mix Adjusted Length of Stay 

1.6.5 PSRO Benefit-Cost Analysis 

1 .6.6 Overview of PSRO Review Costs 

1.6.7 Quality of Care Studies 

1.6.8 Profile Analysis 

1.6.9 Organization Models and 
Performance Assessments 

1.6.10 Program Implementation Status 

1.7 Conclusions 

2. Utilization Impact Studies 

2.1 Overview of Utilization Studies 

2.1.1 Objectives 

2.1.2 Studies and Relationships 

2.2 PSRO Impact on Medicare Hospital 
Utilization Study 

2.2.1 Objectives 

2.2.2 Data Sources 

2.2.3 Methods 

2.2.4 Findings: Descriptive Analyses 

2.2.5 Results: Forced Order Regression 
Analysis 

2.2.6 Results: Impact Analysis 

2.2.7 Discussion of Results 

2.3 PSRO Impact on Specific Diagnoses 
and Procedure Categories Study 

2.3.1 Background 

2.3.2 Literature Review as Basis for 
Procedure and Diagnosis 
Selection 

2.3.3 Methods 

2.3.4 Diagnostic- and Procedure- 
Specific Results 



Page 
ix 

xi 

xvii 

1 
1 
1 
1 
1 
2 
2 
3 
3 
4 

4 
4 
4 



9 
10 
10 
11 
12 
12 

12 
13 
14 

16 
16 
16 
16 

17 
17 
17 
18 
23 

27 
35 
36 

39 
39 



39 
42 

44 



2.4 Medicaid Hospital Utilization Rate Study 

2.4.1 Objectives 

2.4.2 Data and Case Study Area 
Selection 

2.4.3 Issues and Problems in 
Developing Rates 

2.4.4 Methodology for Case Study 
Analysis 

2.4.5 Case Study Findings 

2.4.6 Recommendations 

2.5 Case-Mix Adjusted Length of Stay 

Analysis 

2.5.1 Objectives 

2.5.2 Data 

2.5.3 Methods 

2.5.4 Limitations 

2.5.5 Findings 

2.5.6 Conclusions 

3. Benefit-Cost Analysis of Concurrent Review 

3.1 Objectives 

3.2 Medicare Concurrent Review Costs 

3.3 Medicare Concurrent Review Benefits 

3.3.1 Calculation of Benefits 

3.3.2 Societal Benefits Associated with 
PSRO Medicare Concurrent 
Review 

3.4 Data 

3.5 Findings 

3.5.1 Reimbursement Savings for Four 
Primary Effects 

3.5.2 Long-Term Care and Ambulatory 
Substitution Effects 

3.5.3 Potential Shift of Fixed Costs to 
Non-Medicare Patients 

3.5.4 Benefit-Cost Analysis Results 

4. Overview of PSRO Review Costs 

4.1 Objectives 

4.2 Data Sources and Cost Definitions 

4.3 Findings 

5. Quality Assurance Activities: Medical Care 
Evaluation Studies 

5.1 Objectives 

5.2 Definitions of MCEs 

5.3 Evolution of PSRO Quality Assurance 
Methods 

5.4 Development and Current Status of 
MCE Activity 

5.4.1 Objectives 

5.4.2 Data 

5.4.3 Methods 

5.4.4 Quantity and Distribution of MCE 
Activity 



Page 

60 
60 

61 

61 

65 
67 
72 

73 
73 
74 
74 
75 
76 
78 

81 
81 
81 
82 
82 



83 
84 
84 

84 

87 

87 
87 

89 
89 
89 
89 



93 
93 

93 

93 

94 
94 
95 
95 

95 



5.5 Variation Rate Study 

5.5.1 Objectives 

5.5.2 Study Design and Data Collection 

5.5.3 Methods of Analysis 

5.5.4 Limitations 

5.5.5 Findings 

5.6 Benefits and Costs of Medical Care 
Evaluation Studies 

5.6.1 Objectives 

5.6.2 Methods 

5.6.3 Findings 

5.6.4 Discussion 

6. Profile Analysis 

6.1 The Mandate for Profile Analysis 

6.2 The Nature of Profile Analysis 

6.3 Program Development 

6.4 Description of Profile Analysis Activity 

6.4.1 Data 

6.4.2 Findings 

6.5 Some Characteristics of PSROs that Uses 
Data Effectively 

6.5.1 Recognition of Potential Uses 

6.5.2 Commitment to Address Problems 

6.5.3 Analysis and Implementation 
Capability: Organization of PSRO 
Staff and Physicians 

6.5.4 Data Quality 

6.5.5 Interactive and Flexible Data 
Systems 

7. Organizational Effectiveness: Models and 
Measures for Assessing PSRO Effectiveness 

7.1 Introduction 
7.1.1 



7.2 



Overview of Two General Models 
of Organizational Effectiveness 
The Goal Model of Organizational 
Effectiveness 
The Systems Model of 
Organizational Effectiveness 
Distinctions Between Goal 
Attainment and Systems Models 
The Nature of PSROs 

7.2.1 Formalized System 

7.2.2 Externally Mandated System 
Externally Authorized System 
Regulatory System 
Local Physician Organization 
Performance Evaluations 
Financial and Professional 
Sanctions 



7.1.2 



7.1.3 



7.1.4 



7.2.3 
7.2.4 
7.2.5 
7.2.6 
7.2.7 



Page 
108 
108 
108 
115 
116 
117 

121 
121 
121 
124 
124 

128 
128 
128 
131 
131 
131 
132 

135 
135 
136 



136 
137 

137 



139 



140 



7.2.8 Summary: The Nature of PSROs 

7.3 Observations and Postulates to be 
Tested about Factors Affecting the 
Effectiveness of PSROs 

7.3.1 From the Perspective of the 
Goal Model 

7.3.2 From the Perspective of the 
Systems Model 

7.3.3 Implications 

7.4 Applicability of the Goal and Systems 
Models for Assessing PSRO 
Effectiveness 

7.4.1 The Goal Model 

7.4.2 The Systems Model 

7.4.3 Setting Realistic Performance 
Standards 

1979 Program Status and Directions 

8.1 Introduction 

8.2 Implementation Status 
8.2.1 Program Funding 

Physician Membership 

Review Implementation 

Area Redesignation 

Data Implementation 

National Professional Standards 

Review Council 

Statewide Professional Standards 

Review Councils 

8.3 Program Components and Management 
Initiatives 

8.3.1 Background 

Implementation Initiatives 
Termination of Funding 
General Management Initiatives 
Hospital Review Implementation 
Other Forms of Review 

8.4 Analysis of 1978 PSRO Performance 
Ratings 

8.4.1 Bacl<ground 

8.4.2 Objective 
Methodology 
Data Limitations 
Findings 

PSRO Performance and Utilization 



8.2.2 
8.2.3 
8.2.4 
8.2.5 
8.2.6 

8.2.7 



139 


8.3.2 


139 


8.3.3 




8.3.4 


139 


8.3.5 




8.3.6 



141 


8.4.3 


141 


8.4.4 


141 


8.4.5 


142 


8.5 PSRC 


142 


8.6 Cone 


142 




142 


Appendix 1: 


142 


Appendix II: 



142 



PSRO Implementation Status 
Glossary of Abbreviations and 
Acronyms Used in this Volume 
Appendix III: Glossary of Terms 



Page 

143 

143 

143 

144 
145 



146 
146 
149 

151 

152 
152 
152 
152 
154 
154 
154 
157 

158 

158 

158 
158 
158 
159 
159 
168 
176 

178 
178 
179 
179 
180 
181 
185 
185 

187 
192 

195 



TABLES 



Table 

1 
2 



10 



11 



12 



13 



14 



15 



16 



17 



Components of the 1979 Evaluation Study 

Estimated PSRO impact on Days of Care 

per 1000 Aged IVIedicare Beneficiaries, 

by Region 

Unadjusted Vs. Adjusted Indices, 

12 PSROs 

Benefit-Cost Ratios for Three Alternative 

Assumptions Concerning Ambulatory 

Substitution Effects 

National Summary of PSRO Review Cost 

Components: A Comparison of Results 

from the 1978 and 1979 Evaluations 

Total Days of Care per 1000, Total 

Discharges per 1000 and Average 

Length of Stay for Aged Medicare 

Beneficiaries by PSRO Area, 1974 to 

1978: U.S. Totals 

Total Days of Care per 1000, Total 

Discharges per 1000 and Average Length 

of Stay for Aged Medicare Beneficiaries 

by Active and Inactive PSRO Areas, 1974 

to 1978: Northeast 

Total Days of Care per 1000, Total 

Discharges per 1000 and Average Length 

of Stay for Aged Medicare Beneficiaries 

by Active and Inactive PSRO Areas, 1974 

to 1978: North Central 

Total Days of Care per 1000, Total 

Discharges per 1000 and Average Length 

of Stay for Aged Medicare Beneficiaries 

by Active and Inactive PSRO Areas, 1974 

to 1978: South 

Total Days of Care per 1000, Total 

Discharges per 1000 and Average Length 

of Stay for Aged Medicare Beneficiaries 

by Active and Inactive PSRO Areas for 

Years 1974 to 1978: West 

Mean Values of Adjustment Variables for 

Active and Inactive PSRO Areas 

Comparison of Adjustment Variables 

Between Active and Inactive PSROs, 

by Region 

DOC (1978) Summary of Proportional 

Increments to Explained Variance 

DOC (1978) Summary of Proportional 

Increments to Explained Variance: 

Northeast Region 

DOC (1978) Summary of Proportional 

Increments to Explained Variance: North 

Central Region 

DOC (1978) Summary of Proportional 

Increments to Explained Variance: South 

Region 

DOC (1978) Summary of Proportional 

Increments to Explained Variance: West 

Region 



Page 

7 


Table 

18 


8 


19 


10 





11 



12 



23 



24 



25 



25 



30 



31 



31 



20 

21 

22 

. 23 

24 
25 

26 
27 
28 
29 



25 




26 






30 


28 




29 


31 


30 


32 




33 



34 

35 
36 



Page 

Summary of Proportional Increments to 
Explained Variance for DISC and ALOS 
Regressions: Northeast 32 

Summary of Proportional Increments to 
Explained Variance for DISC and ALOS 
Regressions: North Central 33 

Summary of Proportional Increments to 
Explained Variance for DISC and ALOS 
Regressions: South 34 

Summary of Proportional Increments to 
Explained Variance for DISC and ALOS 
Regressions: West 35 

Unstandardized Regression Coefficients 
for Prediction Model of Total Days of 
Care 37 

Estimated PSRO Impact on Days of Care 
per 1000 Aged Medicare Beneficiaries, 
by Region 38 

Estimated Percent Reduction in DOC: 
1 978 and 1 979 Study Results 38 

Acute Myocardial Infarction Average 
Length of Hospital Stay for Aged 
Medicare Beneficiaries for 1973 and 1977, 
by Active and Inactive PSROs 44 

Discharges per 1000 Aged Medicare 
Female Beneficiaries, for 1973 and 1977, 
for Active and Inactive PSROs 47 

Cataract Surgery Discharges per 1000 
Aged Medicare Beneficiaries, for 1973 
and 1977, by Active and Inactive PSROs 52 

Cholecystectomy Discharges per 1000 
Aged Medicare Beneficiaries, for 1973 
and 1977— Active and Inactive PSROs 52 

Ratio of Radical Mastectomies to all 
Mastectomies per 1000 Aged Female 
Medicare Beneficiaries, for 1973 and 
1977— Active and Inactive PSROs 55 

Ratio of Radical Mastectomies to all 
Mastectomies per 1000 Aged Female 
Medicare Beneficiaries in 1973 and 1977 
by Active PSROs, Total United States 59 

Comparison of Statistical Significance 
Between Hypothesized Impact and Non- 
Impact Groups 59 
Medicaid Rate Study Objectives 60 
Summary of Key Issues in Medicaid Rate 
Construction 62 
Diagnosis Categories by Code: H-ICDA-2 
and ICDA-8 66 
Description of Case Study Areas 68 
PSRO 1 : Quarterly Counts of Medicaid 
Eligibles, Discharges, Days of Care, 
ALOS and Rates per Thousand Eligibles, 
1977-1978 68 



III 



Table 

37 PSRO 2: Quarterly Counts of Medicaid 
Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, 
1976-1978 

38 PSRO 3: Quarterly Counts of Medicaid 
Eligibles, Discharges, Days of Care, 
ALOS and Rates per Thousand Eligibles, 
1976-1978 

39 PSRO 4: Quarterly Counts of Medicaid 
Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, 
1976-1978 

40 Findings of Quarterly Trend Analysis, 
1976-1978 

41 PHDDS Data Elements 

42 Unadjusted ALOS, 12 PSROs 

43 Unadjusted Vs. Adjusted Indices, 12 
PSROs 

44 Unadjusted Vs. Adjusted Indices, 12 
PSROs by Beneficiary Group 

45 Medicare Review Costs in 1978 

46 Sources of Savings Attributed to 
Medicare Reimbursement Formula 

47 Long-Term Care and Ambulatory 
Substitution Effects 

48 Maximum Potential Shift of Fixed Costs 
to Non-Medicare Patients, Hospitals, and 
and Other Payers 

49 Benefit-Cost Ratios for Three Alternative 
Assumptions Concerning Ambulatory 
Substitution Effects 

50 Cost Data Included in Unit Cost Report 

51 Summary of 1978 PSRO Review Cost 
Components 

52 Distribution of PSROs Among Total 
Review Cost per Discharge Categories 

53 Distribution of PSROs Among Concurrent 
Review Cost per Discharge Categories 
by Delegation Status 

54 Median Concurrent Review Costs by Size 
of PSRO and Delegation Status 

55 Median Concurrent Review Costs by 
Months by Review Activity and 
Delegation Status 

56 Quarterly Distribution of MCE Audit 
Activity 

57 Quarterly Distribution of MCE Reaudit 
Activity 

58 The Ten Most Frequently Audited Medical 
Topic Categories, 1978 by Quarter 

59 The Ten Most Frequently Audited 
Surgical Topic Categories, 1978 by 
Quarter 

60 The Ten Most Frequently Reaudited 
Topic Categories, 1978 

61 Criteria for Medical Topics 

62 Criteria for Surgical Topics 

63 MCE Sample Count by Topic 

64 Response Rates 



Page 



69 



70 



70 



Table 

65 

66 
67 
68 

69 

70 



71 




74 


71 


76 




76 


72 


78 




61 


73 


85 


74 




75 


86 


76 




77 




78 


88 


79 




80 




81 


88 


82 


90 


83 


90 


84 


91 


85 




86 


91 


87 


92 






88 




89 


92 


90 


99 


91 


105 






92 


107 





107 

107 
110 
112 
113 
114 



93 



94 



95 



Summary of Variation Rate Changes — All 

Topics — All Criteria 

All Studies, Significant Results by 

Criterion Type and Topic 

Summary of Significant Comparisons by 

Criterion Type — All Studies 

Summary of Significant Comparisons by 

Topic — All Studies 

Summary of Variation Rate Changes — 

All Topics — Criteria with Initial Variation 

Rates 10 Percent or More 

Criteria with Initial Variation Rates of 10 

Percent or More: Significant Results by 

Criterion Type and Topic 

Summary of Significant Comparisons by 

Criterion Type — Criteria with Initial 

Variation Rates of 10 Percent or More 

Summary of Significant Comparisons by 

Topic — Criteria with Initial Variation 

Rates of 10 Percent or More 

Summary of Cost and Benefit 

Calculations 

Form HCFA 141 Data Elements 

PSROs Active in Profiling, by Age 

PSROs Active in Profiling, by Size 

Data Sources Used for Profiles 

Types of Profiles 

Stated Reasons for Profiles 

Actions Reported as a Result of Profiles 

Reported Impacts of Profile Analysis 

PSRO Program Funding 

FY 1979 Funding Assigned for the Ten 

PSRO Regions 

Summary of Funded PSRO Physician 

Membership 

PSRO Review Implementation Activity 

by DHEW Region 

Distribution of Physician Membership 

Percentage of Hospitals Participating in 

Medicare Program and Physician 

Assignment Rates Based on Charges 

Number of PSRO Areas, by State 

Implementation Status 

Summary of PSRO Submission Impact 

Objectives 

Summary of Proposed PSRO Impact 

Objectives — Source of Baseline Data by 

Region 

Summary of Proposed PSRO Impact 

Objectives — Methodology and Source 

of Study Data by Region 

Summary of Proposed PSRO Impact 

Objectives — Focus of Objectives by 

Region 

Summary of Proposed PSRO Impact 

Objectives Related to National 

Objectives by Region 

Average Unit Review Cost Assigned to 

Each Region 



Page 

118 

119 
119 
120 

121 

122 

122 

123 

126 
132 
132 
133 
133 
133 
134 
135 
135 
152 

153 

154 

156 
156 



157 
157 
159 

160 



161 

162 

163 

163 
164 



IV 



Table 

96 

97 

98 

99 
100 
101 

102 

103 

104 

105 



Summary of Reported Administrative 

Agreements by Region 

State Medicaid Agencies with Approved 

Monitoring Plans from PSROs 

Number of Signed MOUs Between HSAs 

and PSROs 

Focused Review Activity 

Summary of Ancillary Service Reviews 

Summary of Ancillary Service Review 

Activity, by Region 

Summary of Ancillary Service Review 

Activity, by Methodology 

Summary of Ancillary Service Review 

Activity, by Region and Methodology 

PSROs Funded for Physician Services 

Review 

Ranges and Average Scores for All 

PSROs (109) for Eight Questionnaire 

Components 



Page 


Table 




106 


165 






107 


167 






108 


168 




171 . 


109 


177 




178 


110 


178 


111 




112 


179 






113 


179 






114 




115 



181 



Page 

Distribution of PSROs by Percentage 

of Area Physicians Belonging to PSRO 182 

Distribution of PSROs by Level of 

Physician Involvement 182 

Level of Physician Participation by Level 

of Physician Membership in PSRO 182 

Ratings of Competency of PSRO 

Executive Leadership and Administrative 

Staff 1 82 

Competency Compared to Stability of 

Executive Directors 183 

Competency Compared to Stability of 

Administrative Staff 183 

Rating of PSRO Goal Setting and Success 

in Meeting Goals 183 

Setting of Goals and Objectives Related 

to Success in Meeting Them 184 

Description of Three Selected PSROs 184 

Average Organizational Effectiveness 

Scores by HEW Region 185 



FIGURES 



Figures 

1 



Page 



10 



Total Days of Short-Stay Hospital Care 
per 1000 Aged Medicare Beneficiaries, by 
Active and Inactive PSROs, for Years 
1974 Through 1978: Total United States 
Total Days of Short-Stay Hospital Care 
per 1000 Aged Medicare Beneficiaries, 
by Active and Inactive PSROs, for Years 
1974 Through 1978 
Average Length of Hospital Stay for 
Aged Medicare Beneficiaries for 1973 
and 1977, by Active and Inactive PSROs 
— Acute Myocardial Infarction 
Average Length of Hospital Stay for 
Aged Medicare Beneficiaries for 1973 
and 1977, by Active and Inactive PSROs 
— Acute Myocardial Infarction by Region 
Hysterectomy Discharges per 1000 Aged 
Medicare Female Beneficiaries for 1973 
and 1977, for Active and Inactive PSROs 
Hysterectomy Discharges per 1000 Aged 
Medicare Female Beneficiaries for 1973 
and 1977, for Active and Inactive PSROs 
by Region 

Cataract Surgery Discharges per 1000 
Aged Medicare Beneficiaries for 1973 
and 1977, By Active and Inactive PSROs 
Cataract Surgery Discharges per 1000 
Aged Medicare Beneficiaries for 1973 
and 1977, By Active and Inactive PSROs 
by Region 

Cholecystectomy Discharges per 1000 
Aged Medicare Beneficiaries for 1973 
and 1977, By Active and Inactive PSROs 
Cholecystectomy Discharges per 1000 
Aged Medicare Beneficiaries for 1973 
and 1977, By Active and Inactive PSROs 
by Region 



45 



46 



48 



49 



50 



51 



53 



54 



Figures Page 

11 Ratio of Radical Mastectomies to all 
Mastectomies per 1000 Aged Female 
Medicare Beneficiaries for 1973 and 
1977, By Active and Inactive PSROs 56 

Ratio of Radical Mastectomies to all 
Mastectomies per 1000 Aged Female 
Beneficiaries for 1973 and 1977, By 
Active and Inactive PSROs by Region 57 

Ratio of Radical Mastectomies to all 
Mastectomies per 1000 Female 
Beneficiaries for 1973 and 1977, By 
Active and Inactive PSROs By PSRO Age 58 
Unadjusted vs. Adjusted Indices, 12 
PSROs (All Patients Combined) 77 

Unadjusted vs. Adjusted Indices, 12 
PSROs (Medicare) 79 

Unadjusted vs. Adjusted Indices, 12 
PSROs (Medicaid) 80 

MCE Initial Audit Activity by Quarter 96 

Average PSRO Audit Activity by Quarter 97 

Percent of Initial Audit Activity by 
Ranked Groupings (1977-1978) 98 

MCE Reaudit Activity by Quarter 102 

Average PSRO Reaudit Activity by 
Quarter 103 

Percent of Total Reaudit Activity by 
Ranked Groupings 104 

Overview of Benefit-Cost Strategy « 123 

Health Status Months in Five Care 
Categories: Mean Estimates of 
Physician Panel 125 

Physician Membership 155 

Proposed PSRO Quality Review Policy 173 



12 



13 



14 

15 

16 

17 
18 
19 

20 
21 

22 

23 
24 



25 
26 



VII 



FOREWORD 



This report presents the 1979 evaluation of the 
Professional Standards Review Organization (PSRO) 
program. The evaluation describes the implementation 
of the PSRO program through 1979, but analyzes 
the impact of the program through 1978. The PSRO 
program is administered by the Health Care Financing 
Administration through the Health Standards and 
Quality Bureau. The evaluation was conducted by the 
Office of Research, Demonstrations and Statistics, 
which is a part of HGFA, but organizationally 
independent from the operating Bureau. The evaluation 
report is the third in a series, and shows that, for the 
second consecutive year, PSROs have reduced 
Medicare hospital utilization relative to inactive PSRO 
areas, and that, overall, PSRO Medicare concurrent 
review activity continues to pay for itself. 

In 1977, it was estimated that PSROs saved the 
Medicare program $5 million more than it cost to 
administer the concurrent review of Medicare bene- 
ficiaries. In 1978, the estimate of PSRO savings is 
substantially greater; an estimated $21 million over 
administrative costs. 

The major utilization study presented in this report 
demonstrates, for reasons which are not yet fully 
understood, that PSRO impact is not uniform across 
the nation. We will continue to work to identify the 
factors that contribute to PSRO effectiveness. 

This evaluation report presents the first major study 
of PSRO impact on physician compliance with quality 
of care criteria. The study shows that Medical Care 
Evaluation (MCE) studies can and do identify variations 
from quality care. When PSROs identify variations 
from accepted standards, compliance with those 
standards improves significantly over time. Preliminary 
work on the benefit-cost analysis of MCEs indicates 
the potential for considerable savings resulting from 
improvements in quality of care. 



A second utilization study was conducted which 
analyzed procedures and diagnoses for which PSRO 
impact seemed most likely. In four out of the five test 
cases there was evidence of PSRO impact. 

An attempt was made in this year's evaluation to 
compute Medicaid hospital utilization rates using four 
PSROs as case studies. The analyses indicate that 
existing Medicaid data sources are not sufficiently 
well developed to permit an adequate assessment of 
PSRO activities in that sector. We are continuing to 
work with States and PSROs to produce more refined 
data files for analysis of PSRO impact on hospital 
utilization. 

Although the program is estimated to pay for itself, 
the operational costs are still increasing, partly due 
to the increased conduct of profile analysis, areawide 
MCE studies, and other forms of review activities. 
The program has been working toward a target unit 
review cost of $8.70 per discharge for the end of 
fiscal year 1979. The data indicate that more remains 
to be done in this vital area. PSROs are beginning 
to intensify their focused review activities; this is 
expected to reduce costs further. 

HCFA appreciates the cooperation of the PSROs 
with these evaluation efforts, especially in the collec- 
tion of MCE data. Without seeking to minimize the 
contribution of others, we particularly wish to express 
our appreciation for the cooperation of the physicians 
and nurses who have worked, through the PSRO 
program, to maintain and improve quality of care 
and to help contain the nation's health care costs. 



Leonard D. Schaeffer 
Administrator 

Health Care Financing Administration 
Department of Health, Education, and 
Welfare 



EXECUTIVE SUMMARY 



This report is the third in a series of annual reports 
prepared by the Department of Health, Education, and 
Welfare to assess the implementation and effectiveness 
of the Professional Standards Review Organization 
(PSRO) program, which was created by P.L. 92-603. 
The evaluation, as mandated in the Medicare- 
Medicaid Anti-Fraud and Abuse Amendments of 1977 
(P.L. 95-142), was conducted by the Office of 
Research, Demonstrations, and Statistics (ORDS), 
Health Care Financing Administration (HCFA). The 
1979 PSRO evaluation is an extension of the 1977 
and 1978 evaluations in that its goals and objectives 
are drawn from the Departmental evaluation plan 
which was approved by the National Professional 
Standards Review Council in 1975. The areas of study 
include: 

PSRO impact on hospital utilization 

Benefit-cost analysis of concurrent review 

PSRO review costs 

PSRO impact on quality of care 

PSRO organizational models 

PSRO program status 

Four utilization studies are presented — two 
evaluative and two developmental. The two evaluative 
studies focus upon the Medicare population. The first 
study, an analysis of PSRO impact on hospital 
utilization, employs essentially the same model used 
in the previous two evaluations. Regional analyses 
are emphasized, however, because PSRO impact 
varies significantly across regions. Hospitals included 
in the data base have been revised this year to reflect 
concerns raised by GAO during the review of last 
year's evaluation. An additional year of PSRO experi- 
ence lends added longitudinal perspective to the 
findings. The second evaluation study, involving 
diagnosis- and procedure-specific utilization analyses, 
is more complete than last year's and for the first 
time indicates varying impact across selected 
diagnoses and procedures. 

A developmental Medicaid utilization study provides 
insight into the estimation of Medicaid hospital 
utilization rates. Medicaid rate trends, based on PSRO 
Hospital Discharge Data Set (PHDDS) data and State 
Medicaid eligibility files, are presented for four PSRO 
case study areas. The final utHization study presents 
a developmental analysis of changes in case-mix 
severity and case-mix adjusted lengths of stay, both 
over time and across PSROs. 

This year's benefit-cost analysis of concurrent 
review develops individual benefit and offset com- 
ponents in detail. Net benefits are then compared to 
the costs of concurrent review for Medicare bene- 
ficiaries. While the focus of the benefit-cost analysis 
is directly on savings in Medicare program expendi- 
tures which are related to concurrent review activities, 
the analysis also discusses the impact of the PSRO 
program on non-Federal patients. 



The quality of care section of this year's report is 
considerably expanded. The quality study now repre- 
sents a finished work rather than a pilot study, and 
a Medical Care Evaluation (MCE) benefit-cost 
methodology has been developed and tested on a 
limited data base. A discussion of profile analysis 
augments the discussion of MCEs. An addition to this 
year's report is an organizational analysis of the 
PSRO program which provides insight into the 
structure of the program and how the program fits 
into the broader context of the health care sector. The 
report closes with an update of PSRO program status. 

The key findings of the 1979 PSRO Evaluation, and 
the context within which they were obtained, are 
summarized below: 

Medicare Impact Study (100% Medicare Claims 
File Study): The purpose of the Medicare Impact Study 
is to determine the impact the PSRO program has had 
in reducing inpatient hospital utilization by Medicare 
beneficiaries. The study is based on an analysis 
which compares data from 108 active PSROs (defined 
as PSROs with at least one hospital under review as 
of July, 1978) with that from 81 inactive PSROs over 
the period 1974 to 1978. The primary measure of 
utilization impact is the estimated reduction in total 
days of care per 1,000 aged Medicare beneficiaries 
associated with PSRO activity as of 1978. Regression 
techniques are used to test for PSRO impact. The 
approach used is conservative in that the PSRO effect 
is tested after controlling for the effect of numerous 
demographic and health care supply variables. 

As in last year's evaluation, current results indicate 
that the PSRO program had a small impact on hospital 
days of care per 1,000 aged Medicare beneficiaries. 
While the estimated PSRO effect is in the expected 
direction, it does not meet standard levels of statistical 
significance. However, a statistically significant 
relationship between PSRO impact and the four 
census regions — Northeast, North Central, West, and 
South — was demonstrated. This finding suggests that 
a significant PSRO impact exists but that it varies by 
region, presumably because of factors not yet fully 
understood associated with the regions. Because of 
the relationship between regions and PSRO impact, 
PSRO impacts were estimated separately for each 
region. 

The results of the regional analyses show that PSRO 
areas in the Northeast and North Central regions have 
a statistically significant impact in reducing the 
Medicare days of care rate. The PSRO impact in the 
West region is also in the direction of reducing days 
of care but does not attain statistical significance. In 
the South region the analysis indicates that PSRO 
activity is apparently associated with an increase in 
the days of care rate. 

The Medicare Impact Study utilizes the regression 
results to estimate the number of Medicare "days 
saved" due to PSRO review activity in 1978. The 



XI 



number of days saved was estimated for each region. 
The national estimate is calculated by summing the 
regional days saved. The estimated number of 
Medicare days saved for 1978 is 948,430 days, 
representing 1.7 percent of Medicare days of hospital 
care in active PSRO areas. When approximately the 
same techniques are applied to 1977 data, a 2 percent 
savings is estimated. The estimated relative reductions 
in days of care by U. S. Census region for 1978 are: 

Northeast : —4.8% 

North Central: —2.1% 

South : +3.7% 

West : —1.4% 

U. S. Total : —1.7% 
It is important to stress that the results of the 
Medicare Impact Study apply to the program as 
implemented in 1978. These findings may not be 
legitimately extended to indicate the impact of a fully 
implemented program. In particular, if one were to 
extend the study results to 1979, the unfavorable 
findings estimated for the South would be assigned 
a greater weight since the South was the least 
implemented of the four regions in 1978. Such an 
extrapolation of the unfavorable results in the South 
may not be appropriate. A comparison of adjustment 
variables across active and inactive areas in the South 
suggests that the inactive PSRO areas are highly 
dissimilar from the active areas. Thus, it is not clear 
that an expansion of the PSRO progam in the South 
will continue to produce seemingly unfavorable 
results. 

Although the Medicare Impact Study results 
represent the best estimate of PSRO impact as of 
1978, certain caveats must be kept in mind. First, bias 
may exist because PSROs were not implemented 
randomly, but on a self-selected basis. Second, the 
study examines Medicare data only and does not 
estimate PSRO savings for other Federal beneficiaries 
(Medicaid and Maternal and Child Health). Third, the 
study estimates the impact of PSRO concurrent review 
as compared to existing utilization review, rather 
than against a situation where no review exists. 
Because some utilization review is being performed 
in inactive areas, estimated PSRO impact is not 
representative of absolute savings in days of care but 
rather the incremental days of care savings relative 
to non-PSRO utilization review. Finally, estimated 
PSRO impact represents average savings; it was not 
possible to determine the effectiveness of the best 
PSROs. Organizational characteristics which can 
predict PSRO effectiveness have yet to be determined. 

Diagnosis- and Procedure-Specific Impact Study (20 
Percent Medicare Discharge File): The second analysis 
of Medicare hospital utilization rates tests a priori 
hypotheses that PSRO intervention is or is not likely 
to reduce the utilization rates of selected diagnoses 
and procedures. The methodology used in the 1978 
PSRO Evaluation analysis of Medicare utilization rates 
is utilized because the diagnosis- and procedure- 
specific data base contains data through 1977, as did 
the 1978 PSRO evaluation data base. 



As a result of a medical literature review and 
consultation with experts in the field, four procedures 
and one diagnosis were selected as possiljle areas 
where PSRO impact might be expected. The diagnosis 
is acute myocardial infarction, where it was hypothe- 
sized that PSROs reduce the average length of stay. 
The four procedures are cholecystectomy, cataract 
removal, and hysterectomy, where it was hypothesized 
that PSROs reduce the discharge rate, and breast 
cancer, where it was hypothesized that PSROs reduce 
the proportion of radical mastectomies to total 
mastectomies. 

Study results indicate statistically significant PSRO 
effects for four of the five diagnoses and procedures 
tested. As hypothesized, active PSROs yield a greater 
decline in average length of stay for acute myocardial 
infarction, and in discharge rate for cholecystectomy. 
The cataract surgery discharge rate increased for 
both active and inactive PSROs, but the rate of 
increase was significantly lower for active PSROs, a 
finding consistent with the hypothesis. Finally, there 
is a statistically significant PSRO effect on the use 
of radical mastectomies as a percentage of total 
mastectomies from 1973 to 1977. While younger 
PSROs show an increase in the discharge rate, the 
older ones show a decrease, producing an overall 
favorable impact. Only in the case of hysterectomy 
is there no significant PSRO effect. 

These results must be carefully interpreted. First, 
although the literature and expert testimony suggested 
that selected areas could be affected by PSRO review, 
this does not necessarily imply that individual PSROs 
actually addressed these areas. Second, a study by 
the Institute of Medicine indicates that for specific 
diagnoses and procedures, there exist inaccuracies 
in the coding of the data, which might affect the 
results reported above. Third, the breast cancer 
results are the most tentative; although the relative 
effectiveness of radical mastectomies has been 
questioned for some time in many areas of the 
country, radical mastectomy was still the treatment 
of choice in 1977. 

Medicaid Hospital Utilization Rate Study: There has 
never been a national analysis of the impact of PSRO 
on Medicaid hospital utilization. This is due to the 
fact that there are no national statistics for Medicaid 
hospital utilization or eligibility. The Medicaid Hospital 
Utilization Rate Study explores the potential use of 
two existing data sources, the PSRO PHDDS and 
Medicaid State Agency eligibility data, to develop 
hospital utilization rates. Rate estimation methods are 
developed and applied to data representing four case 
study PSROs. 

Days of care and discharge rates per 1,000 Medicaid 
eligibles and average length of stay are studied. 
Hypotheses relating to changes in these measures 
are tested for three groups: (1) all Federal discharges; 
(2) twenty-two selected diagnostic groups; and (3) 
four surgical status groups. 

Results from analyses of days of care rate trends 
are mixed — two PSROs exhibited declining patterns; 



XII 



two PSROs showed increasing patterns. None of the 
days of care rate changes, however, is statistically 
significant. As observed in the prior PSRO evaluations, 
increases (decreases) in days of care rates are 
associated with increases (decreases) in discharge 
rates. Analyses for specific diagnostic groups exhibit 
a similar mixed pattern. However, both discharge 
and days of care rates for several mental disorders 
show increases in all four case study areas. Finally, 
the results did not establish significant utilization 
trends nor do they permit an assessment of PSRO 
impact (since data are not available for non-PSROs 
were available for comparisons). 

The study analyzes the many problems associated 
with construction of Medicaid utilization rates and 
concludes that Federal evaluation based on Medicaid 
data will eventually require individual unit record 
data. Such data are required to provide adequate 
flexibility to standardize data and analyses across 
individual State Medicaid programs. 

Case-Mix Adjusted Length of Stay Study: This study 
presents a developmental analysis which focuses 
upon the development of case-mix adjusted average 
length of stay (ALOS) measures and case-mix indices. 

The analysis makes use of the PSRO Hospital 
Discharge Data Set (PHDDS) to produce two indices 
which relate directly to ALOS. The first index is a 
measure of ALOS which adjusts for variations in 
case-mix; the second index measures variations in 
case-mix proportions, as reflected by changes in the 
patient age distribution, and the presence or absence 
of multiple diagnoses and surgical procedures. 

These indices are applied to twelve PSROs for 
which PHDDS data tapes of acceptable quality are 
available for all of 1977 and 1978. The results show 
an aggregate increase in case-mix severity over time, 
and an aggregate decrease in case-mix adjusted 
length of stay. Whereas the unadjusted ALOS 
declined by about three percent over the two-year 
period, the case-mix adjusted ALOS declined by about 
twice that amount. Correspondingly, the case-mix 
index increased by about three percent. These effects 
were somewhat more pronounced for Medicare as 
opposed to Medicaid beneficiaries. 

Case-mix adjustments, applied cross-sectionally, 
show that differences in unadjusted ALOS among 
PSROs can be factored into two components: one 
component attributable to differences in case-mix, the 
other to differences in true ALOS behavior. 

PSRO Benefit-Cost Analysis of Concurrent Review 

The benefit-cost analysis estimates the dollar value 
of PSRO-related reductions in days of care for 
Medicare patients and relates these dollar benefits to 
PSRO concurrent review costs for Medicare hospital 
admissions. The resulting benefit-cost ratios indicate 
the number of dollars "saved" for each dollar spent 
for Medicare concurrent review. Benefits are limited 
to changes in reimbursements associated with 
Medicare concurrent review; i.e., benefits resulting 



from review of Medicaid and Title V patients are 
excluded, as are the benefits of other forms of 
Medicare review such as areawide or MCE review. 
The data pertain to 1978 when the PSRO program 
was not fully implemented. 

Benefits are calculated by using the estimated 
percent reduction in Medicare days from the Medicare 
Impact Study and hospital-specific cost data for 1978. 
There are four benefit components corresponding to 
the four types of reimbursement adjustments that occur 
within the Medicare reimbursement formula. Potential 
gross savings are calculated at the hospital level, and 
then aggregated to Census region and national levels. 
This corresponds to the method used to calculate 
the number of days saved. 

Gross savings are then reduced to account for the 
costs that would be generated if long-term care (LTC) 
days were substituted for inpatient short-stay hospital 
days "saved" or if outpatient ambulatory care were 
substituted for inpatient ancillary services "saved". 
Since the value used for the ambulatory care 
substitution effect is critical to the benefit calculations, 
a separate set of benefit-cost estimates is produced 
for each of three ambulatory care substitution 
assumptions (.30, .50, and .70). The .50 ambulatory 
substitution rate, which was used last year, gives a 
national benefit-cost ratio of 1.269 for 1978 PSRO 
Medicare concurrent review activities. Corresponding 
regional results are presented below: 

Benefit-Cost 
Ratio 
Northeast : 3.88 
North Central: 1.79 
South : —1.91 

West : 0.72 



U. S. Total 



1.269 



With an ambulatory substitution rate of .30, the 
benefit-cost ratio reaches 1.504 and at .70 it is 1.035. 
In all three cases, the West has a benefit-cost ratio 
less than one, the positive savings in the North 
Central are slightly more than offset by the negative 
savings in the South, and the Northeast exhibits a 
benefit-cost ratio ranging from 3.17 to 4.58. 

The benefit-cost ratios presented above relate 
incremental benefits to full PSRO concurrent review 
costs. Although the Medicare Impact Study estimates 
benefits which are incremental to utilization review 
(UR) activities, comparable incremental costs could 
not be calculated because reliable estimates of UR 
costs do not exist. If UR benefits are small, the 
approach used is essentially correct since it very 
closely approximates a matching of full benefits to full 
costs. To the extent that UR benefits or costs are 
substantially greater than zero, the results of the 
analysis are biased. 

The focus of the benefit-cost study is on Medicare 
reimbursements. From the societal perspective, 
however, one should take into account the transfer 
of costs to the non-Medicare population by the 
Medicare reimbursement formula. When this issue 



XIII 



is examined it is found that up to half of the estimated 
savings in reimbursements may represent fixed costs 
which are absorbed by other payors. 

In determining the reduction in reimbursement due 
to saved Medicare days, the ratio of variable to fixed 
costs was assumed to be 40:60. However, with a 
longer planning horizon, hospitals will have greater 
discretion over investments and a smaller percentage 
of total costs will be fixed. Also, as cost containment 
becomes a key concern of more government agencies, 
the probability increases that hospital capacity will be 
reduced. Thus, relative to present trends, the supply 
of beds and capital equipment may decrease. To the 
extent that this proves true, 40 percent is probably 
an underestimate of hospital variable costs over 
longer time horizons. From a long run perspective, 
then, using the 40:60 assumption leads to an under- 
estimate of the impact of PSRO review on reductions 
in Medicare reimbursements. 

PSRO Review Costs 

Data on review costs for individual PSROs were 
tabulated from the Unit Cost Report for 1978 prepared 
by the Health Standards and Quality Bureau (HSQB), 
HCFA. Median concurrent and MCE review cost per 
discharge declined between 1977 and 1978 ($8.76 to 
$8.57 for concurrent and $1 .40 to $1 .1 9 for MCE). This 
decline, however, is more than offset by an increase 
in areawide review costs from $2.21 to $3.24. (Area- 
wide review costs include the costs of physician and 
patient profiles, of computer support, and of monitoring 
delegated review functions.) The result is a net 
increase in median total hospital review costs from 
$12.31 to $12.91. Federal management and support 
costs also increased slightly from $5.07 to $5.33. The 
$12.91 figure is considerably greater than the target 
of $8.70 per Federal discharge set for the end of 
FY 1979. 

Since there is no information on the number of 
reviews conducted, there is no way to calculate the 
cost per review. However, the costs per discharge do 
not differ substantially between the small and large 
PSROs. 

PSRO Impact on Quality of Care 

The impact of MCE studies on the quality of care 
provided in PSRO hospitals is examined by collecting 
data from a representative sample of MCE audits and 
reaudits. The measure of quality is the variation rate, 
defined as the proportion of patient records which 
do not meet standards set by physicians. In previous 
PSRO evaluations, the variation rate was validated 
as a measure of change in quality of care, but this 
is the first time it has been applied to a representative 
sample of MCEs. 

The MCE study includes 710 audit and reaudit 
combinations in six medical diagnoses and six surgical 
procedures. Variation rates are calculated at audit and 
at reaudit. Presumably actions were taken prior to 
reaudit to correct problems in care. The principal 



question under study is whether those actions, if 
taken, produced a significant reduction in variation 
rates between audit and reaudit. 

The study results show a statistically significant 
trend. Variation rates at reaudit tend to be lower than 
those at audit, indicating measured improvements in 
quality of care. This is especially true for those varia- 
tion rates greater than 10 percent at initial audit. These 
are the cases which reflect more serious problems in 
care, and which are more likely to be acted upon by 
medical audit committees. For these instances the im- 
provement between audit and reaudit was pronounced. 

The findings of the study must be interpreted in light 
of several limitations. The response rate, while high 
(72%), leaves room for bias associated with the self- 
selection of responding hospitals. Perhaps the most 
important limitation is the problem of attribution. The 
study indicates changes in care between audits and 
reaudits, but cannot conclude that the changes were 
caused by MCEs since no comparison groups were 
available for analysis. 

Variation rate changes in a sample of MCEs are also 
used as a basis for estimating the health benefits re- 
sulting from MCEs. The benefits are related to two 
kinds of costs: the costs of conducting MCEs, and the 
costs of the changes in care brought about as a result. 
Findings indicate that even small improvements (reduc- 
tions) in variation rates may have major implications for 
patient health status, and that when the monetary value 
of improved longevity and functioning are compared to 
MCE costs, benefits could far exceed costs. 

An investigation of profile analysis finds that 123 
PSROs indicated some sort of profile activity in 1979. 
Profile analysis, generally the last component of the 
PSRO review system to be implemented, is receiving 
emphasis from the program as a way of increasing the 
efficiency of review. While PSROs seem to have 
become more accomplished in their use of data, pro- 
filing is not yet supporting concurrent review and other 
review activities except for the most advanced PSROs. 

PSRO Organizational Models 

Past PSRO evaluations have contained little work on 
the organizational characteristics of PSROs and how 
these characteristics might be expected to and do 
affect PSRO effectiveness. This report contains a re- 
view of the research literature relevant to the study of 
PSRO performance. PSROs are described as for- 
malized, externally authorized, and mandated local 
physician organizations expected to function as a 
regulatory system exercising control via performance 
evaluation tied to financial and professional sanctions. 
In their capacity as locally organized physician regu- 
latory systems, the PSRO incorporates features of 
earlier control systems governing the delivery of care 
in hospitals; however, insofar as they are more highly 
formalized, are externally validated and mandated and 
have their decisions backed by financial as well as 
professional sanctions, PSROs represent a new form 
of control over medical practice. 



XIV 



Two important functions of the PSRO should have a 
positive impact on the quality assurance and cost re- 
duction activities in hospitals. First, the information 
gathering and processing activities of the PSRO make 
it, at least potentially, a force for change and reform. 
Gradually, data systems are becoming operational. The 
availability of comparable data sets for all hospitals will 
permit the identification of "outliers," those hospitals 
and physicians whose performance is substantially at 
variance from the standards of care provided by the 
average providers. 

Second, PSROs can have some impact on the per- 
formance of hospitals by the types of training and sup- 
port which they provide for the hospital review co- 
ordinators and physician advisors. Prior to the exist- 
ence of PSROs, utilization review and quality assur- 
ance personnel functioned within the confines of their 
own hospitals, often limited in the types of data which 
they could collect, and lacked information on the per- 



formance levels achieved in comparable hospitals. 
Thus, the PSRO program has brought more uniformity, 
and possibly more leverage, to the utilization review 
process. 

PSRO Program Status 

The program status section provides detailed de- 
scriptive information on the various aspects of the 
PSRO program and the extent to which they are being 
implemented. Of particular relevance for this evalua- 
tion are the tabulations on the extent of implementation 
of HSQB program initiatives designed to reduce pro- 
gram costs and improve management. Although prog- 
ress was made by PSROs toward implementing the 
initiatives instituted in 1978, it was not extensive enough 
for the present mode of operation to be considered 
significantly different from that reflected in prior 
evaluations. 



XV 



CONTRIBUTORS 



The 1979 PSRO evaluation was managed by the 
Office of Research within the Office of Research, 
Demonstrations, and Statistics (ORDS) of the Health 
Care Financing Administration (HCFA). The evaluation 
report is the product of the work of many people. The 
project directors are Allen Dobson, Ph.D. and Feather 
A. Davis, Ph.D. Project staff are Gerald S. Adier, M. 
Phil.; David K. Baugh, M.A.; Paul W. Eggers, Ph.D.; 
Bennett B. Hirsch, M.S.; Helen Kong, B.A.; Roger 
McClung, Ph.D. and Sherry A. Terrell Ph.D. 

The evaluation was aided by a Technical Advisory 
Panel composed of individuals of recognized expertise 
in the area of health care review. The panel members 
were Robert H. Brook M.D., Sc.D.; Linda K. Demlo, 
Ph.D.; Michael D. Intriligator, Ph.D.; Donald G. Riedel, 
Ph.D.; and W. Richard Scott, Ph.D. John W. Bussman, 
M.D. represented the National Professional Standards 
Review Gouncil. 

The design and conduct of the evaluation has been 
reviewed by representatives of the Office of the HEW 
Assistant Secretary for Planning and Evaluation; Office 
of the HEW Assistant Secretary for Management and 
Budget; Health Gare Financing Administration; Office 
of Management and Budget; and the Gongressional 
Budget Office. Special acknowledgment is extended to 
Daniel Koretz, Ph.D. and Paul Ginsburg, Ph.D. of the 
Gongressional Budget Office. 

The hospital utilization studies, coordinated by Paul 
W. Eggers, comprise two evaluative analyses of Medi- 
care claims data and two developmental analyses of 
the PSRO Hospital Discharge Data Set (PHDDS). The 
Medicare Impact Study, which uses the Medicare 100 
Percent Glaims File, was conducted by Paul W. Eggers 
with statistical support provided by G. Mitchell Dayton, 
Ph.D., of the University of Maryland. The analysis of 
the 1973-77 Medicare 20 Percent Research File data 
was conducted by Bennett Hirsch, Sherry A. Terrell, 
Helen Kong, and Paul W. Eggers, with the assistance 
on specification of diagnoses and procedures by 
William Finkle, Ph.D., of the Harvard University School 
of Public Health. Gratitude is expressed to several 
members of ORDS for data preparation and proces- 
sing: David K. Baugh, Marilyn Newton, Ronald Prihoda, 
Mae Robinson and, Mary White. 

The two developmental analyses of the PHDDS data 
were directed by David K. Baugh and Feather A. Davis, 
with contract support from Margaret O'Brien, M.H.S.A., 
Franz Jaggar, B.A., Scot Beam, M.A., Embry Howell, 



M.S.P.H., and Jerilyn Woelfel, M.A., of SysteMatrics, 
Inc. and Leonard Greenberg, M.S. and John Rafferty, 
Ph.D. of JWK International Gorporation. Appreciation is 
also expressed to Sharmin Stevens, R.N., and Pearlena 
Patters, M.A., DPR/HSQB; and Michael Davis, M.A., 
DDPA/HSQB, for assistance with the PHDDS data 
preparation and contract support. 

The cost and benefit-cost studies were coordinated 
by Roger McGlung. Providing contract assistance of 
these studies were Franklin E. Robeson, Ph.D. and 
William Geary, Ph.D. of William and Mary Gollege and 
Roger Pfaffenberger, Ph.D. of Texas Ghristian Univer- 
sity. ORDS staff who assisted in the analyses were 
Peter McMenamin, Ph.D., David K. Baugh, Sherry A. 
Terrell, and David Herring, DDPA/HSQB. 

The analyses of the medical care evaluation studies 
were directed by Gerald AdIer, with contractual assist- 
ance from The Rockburn Institute, Inc. Gontributors 
include Maureen Rothermich, R.N., and John Stewart, 
M.A., of HSQB; Karen Orloff Kaplan, M.S.W., M.P.H., 
and Dale Schumacher, M.D., of Rockburn Institute; 
John Williamson, M.D., Susan Horn, Ph.D., and David 
Salkever, Ph.D., of The Johns Hopkins University; and 
William Jessee, M.D., and Myra Pfisterer, R.N., of Medi- 
cal Gare Evaluation Systems, Inc. 

The profile analyses section was coordinated by 
Gerald AdIer, with contractural assistance from 
Gregory Banks of InterQual, Inc. 

The organizational analyses were coordinated by 
Feather Davis. The theoretical model research was 
conducted by W. Richard Scott, Ph.D., Mitchell La 
Plante and Robert Bies, M.A., of Stanford University. 

The program status section was prepared by Helen 
Kong with assistance from Michael Ney, M.P.A., of 
JWK International, and Michael Davis and numerous 
others from HSQB. The aniysis of the project officer 
performance ratings was performed by contractor 
Bonnie Owen, M.A. 

Appreciation is expressed to the secretarial and 
statistical assistance staff of the Division of Beneficiary 
Studies, ORDS, in particular, Anitha Hoyle, Garole 
Weinberg, Jill Hare, Rosemary Goode, and Kathryn 
Barrett. 

Finally, acknowledgment is made to Judith Lave, 
Ph.D., Director, Office of Research, ORDS, and Leonard 
Greenberg for their valuable substantive and editorial 
commentary. 



XVII 



1. Introduction and Summary 

1.1 Background 

This report is tine tliird in a series of annual reports 
prepared by tine Department of Health, Education and 
Welfare to assess the implementation and effectiveness 
of the Professional Standards Review Organization 
(PSRO) program. The evaluation was conducted by the 
Office of Research, Demonstrations, and Statistics 
(ORDS), Health Care Financing Administration (HCFA). 
As with the previous two reports, this volume is 
intended primarily to serve as an analytical resource 
for the many persons with responsibility for, and 
interest in, the PSRO program. However, some limited 
policy recommendations will be made. 

The 1979 PSRO evualation strategy is an extension 
of the 1977 and 1978 evaluations in that its goals and 
objectives are drawn from the departmental evaluation 
plan which was approved by the National Professional 
Standards Review Council in 1975.^ The areas of 
study include: 

• PSRO impact on hospital utilization 

• Benefit-cost analysis of concurrent review 

• PSRO administrative costs 

• PSRO impact on the quality of hospital care 

• PSRO organizational characteristics 

• PSRO program status 

The major findings from these analyses are 
presented in this summary section without extensive 
discussion of the research methodologies. Sections 
2 through 8 present detailed discussions of the study 
hypotheses, design, analyses and findings. 

1.2 Context 

To further the reader's understanding of both the 
PSRO program and the evaluative questions asked of 
PSROs, this section briefly reviews the context within 
which the PSRO program has evolved. 

1.2.1 The Problems 

The PSRO program legislation was a result of 
Congressional concern with increased absolute health 
care costs as well as the increased proportion of 
Federal and State budgets being spent on health care 
and concern over the quality of care being rendered 
to Federal beneficiaries. Overall, Federal budget 
outlays for health have grown from 6.6 percent of the 
Federal budget in 1970 to 9.7 percent in 1978.^ 
Medicare hospital expenditures increased from 2.3 

^Office of the Assistant Secretary for Health, Office of 
Professional Standards Review, Program Evaluation Plan: 
Professional Standards Review Organizations, September 
22, 1975. 

"Economic Report of ttie President, (GPO: Washington 1979), 
pp. 264-5. The cited percentages exclude the Defense 
Department health budget in the numerator, but include it 
in the denominator. 



billion dollars in 1967 to 17.4 billion dollars in 1978. 
At an average annual rate of increase of 14.5 percent, 
hospital expenditures have been the fastest growing 
expenditure category under Medicare. 

Studies have shown that the current set of financing 
mechanisms has not resulted in an efficient allocation 
of societal resources to and within the hospital indus- 
try. It has been noted that since the Federal and State 
governments became major third party payers of 
health care as a result of the Medicare and Medicaid 
legislation of 1965, hospitals have become assured of 
a steady source of revenue. Hospitals have been 
reimbursed by Medicare and Medicaid, as they have 
been by many Blue Cross plans, for the cost of 
delivering care. "With Blue Cross and the government 
reimbursing hospitals on the basis of incurred costs, 
providers were nearly guaranteed reimbursement for 
whatever they chose to do, whether it was the acquisi- 
tion of new equipment, the addition of new beds, 
performance of more lab tests or the hiring of addi- 
tional personnel. Incentives to economize . . . were 
removed completely." ^ Whether as a consequence or 
concomitantly, hospital costs rose at a rate sub- 
stantially more rapid than that of inflation, chiefly 
through expansion of hospital beds, service availability, 
personnel, and service intensity. One response to this 
rapid growth has been regulation. * 

1.2.2 The Response 

Historically, most third party payers, including 
government agencies, exerted little effort to affect the 
quality or price of the hospital service rendered.' 
Whereas most of the pre-Medicare regulations promul- 
gated by private. State, local and Federal agencies 
were intended to promote quality assurance through 
accreditation and licensure of facilities and personnel, 
the post-Medicare period has seen a proliferation of 
piecemeal legislative and administrative efforts 
designed to contain hospital costs through controls 
on capital expansion, rate structures, and utilization." 

Four major hospital regulatory programs have been 
developed since the passage of Medicare: (1) manda- 
tory health planning (the National Health Planning 
and Resources Development Act of 1974, P.L. 
94-641) which requires State-designated agencies to 
regulate changes in bed size, changes in services 
offered, and the expansion and/or modernization of 
hospital facilities through certificate of need programs 



■"Abt Associates, Inc., National Hospital Rate-Setting Study: 
Research Objectives and Metliodology (September 14, 
1979) under Contract No. HEW 500-78-0036, pp. 1-5, 1-6. 

^ Ibid, pp. 1-6, 1-7. It should perhaps be pointed out that 
not everyone concurs that the growth has been excessive. 

^ A. R. Somers, Hospital Regulation: Dilemma of Public Policy 
(Princeton, 1969), p. 115. 

" F. A. Sloan and B. Stelnwald, "Effects of Regulation on 
Hospital Costs and Inputs Use," Journal of Law and 
Economics, forthcoming, 1980; also see, F. D. Hair, 
"Hospital Accreditation; A Development Study of the 
Social Control of Institutions," (Ph.D. dissertation, 
Vanderbilt University, 1972). 



which review hospital capital expenditures in excess 
of $100,000; (2) the State hospital capital investment 
limits authorized under P.L 92-603 (Section 1122 of 
the Social Security Act) which allow volunteer States 
to use the Medicaid and Medicare reimbursement 
mechanisms to regulate hospital facility and service 
growth; (3) State rate-setting commissions, which 
differ extensively in the way they prospectively review 
hospital budgets and/or set reimbursement rates for 
individual hospitals; ^ and (4) utilization review mecha- 
nisms which were required as a condition of participa- 
tion for hospitals under Medicare. 

Utilization review is required by Titles XVIII and 
XIX of the Social Security Act for the purpose of 
controlling resource utilization by patients through an 
active and ongoing monitoring mechanism. The 
Department of Health, Education, and Welfare 
implemented the intent of the law through the require- 
ment that hospitals and extended care facilities 
submit a plan for operation and that a standing 
medical committee be established for: 

(1 ) "review on a sample or other basis of admission, 
duration of stay and professional services 
furnished and, 

(2) review of each use of continuous extended 
duration." ** 

Utilization review was initially instituted to monitor 
hospital admission and length of stay. As implemented, 
utilization review became a mechanism for medical 
staff education in which peer analysis of quality of 
care was combined with the determination of 
appropriate use of hospital resources." 

Over time, hospital-based utilization review came to 
be viewed as an ineffective utilization control 
mechanism and the concept of a physician-sponsored 
peer review mechanism evolved as a potential 
substitute for utilization review.^" The concept of 
regionalized peer review which emphasized both the 
improvement of quality of medical care and the 
moderation of its cost was developed further in 1970 
through research and demonstration monies from the 
National Center for Health Services Research and 
Development." This interest in peer review culminated 
in the creation of the Professional Standards Review 
Program. 1- Utilization reviews remains a HCFA condi- 



■ Abt Associates, Inc., op. cit. 

■31FR13424, October 18, 1966. 

'Arthur D. Little, Inc., An Evaluation of the Effectiveness of 
Utilization Review Activities In Hospitals and Extended Care 
Facilities (1972), p. 15. 

' B. Decker, and P. Bonner, PSRO: Organization for Regional 
Peer Review. (Ballinger: Cambridge, 1973), pp. 8-24. 

^ National Center for Health Services Research and Develop- 
ment, Experimental l\/ledical Care Review Organization 
(EMCRO) Programs, DHEW Publications No. (HSM) 73-3017, 
March, 1973. 

' USDHEW, PHS, HSA, OPEL, Volume 2: A Context for 
PSRO Evaluation: Tlie Program and Evaluation Precedents 
(October, 1977), No. 77-12. 



tion for Medicare participation for those hospitals 
not yet under PSRO review." 

1.3 The PSRO Program 

1.3.1 Mandate 

In 1972 the peer review concept became law with 
the establishment of the Professional Standards 
Review Organization (PSRO) through Public Law 
92-603. In the seven years since implementation, the 
PSRO legislation has been given form and substance 
by the Department of Health, Education, and Welfare." 
Considerable conflict has surrounded the development 
of PSROs, ranging from active opposition by organized 
medicine to brief budgetary elimination of the program 
by the Office of Management and Budget. 

Some of the controversy about the PSRO program 
is attributable to the dual mandate and the lack of 
specificity contained in the following legislative 
Declaration of Purpose: 

Section 1 1 51 . In order to promote the effective, 
efficient, and economical delivery of health care 
services of pi-oper quality for which payment may 
be made (in whole or in part) under this Act and 
in recognition of the interests of patients, the public, 
practitioners, and providers in improved health care 
services, it is the purpose of this part to assure, 
through the application of suitable procedures of 
professional standards review, that the services for 
which payment may be made under the Social 
Security Act will conform to appropriate professional 
standards for the provision of health care and that 
payment for such services will be made — (1) only 
when, and to the extent, medically necessary, as 
determined in the exercise of reasonable limits of 
professional discretion; and (2) in the case of 
services provided by a hospital or other health care 
facility on an inpatient basis, only when and for 
such period as such services cannot, consistent 
with professionally recognized health care standards, 
effectively be provided on an outpatient basis or 
more economically in an inpatient health care 
facility of a different type, as determined in the 



" Prior to the creation of HCFA, the Departmental respon- 
sibility for the implementation of utilization review was 
shared by the Public Health Service which wrote the 
standards for the Conditions of Participation and surveyed 
the institutions, and the Social Security Administration 
which implemented the Medicare program. Since 1977 
both functions have been the responsibility of HCFA, 
implemented by the Office of Standards and Certification, 
HSQB, and the Bureau of Program Operations. 

"The PSRO Program is administered by the Office of 
Professional Standards Review Organizations, Health 
Standards and Quality Bureau, Health Care Financing 
Administration (HCFA). Prior to the establishment of HCFA 
in 1977, it was administered by the Bureau of Quality 
Assurance, Health Services Administration, U.S. Public 
Health Service. 



exercise of reasonable limits of professional 
discretion. 15 

The legislation thus gave the PSROs responsibility 
for limiting the cost of care while assuring the proper 
quality of care. The relative importance placed on 
these complementary, yet sometimes conflicting, 
objectives differs among the concerned parties. The 
medical profession has stressed the quality assurance 
aspects of the program; the Congress, however, has 
stressed the utilization control mission. 

1.3.2 Implementation 

In order to meet the legislative objectives set for 
the program, DHEW developed a model for acute care 
hospital review which incorporated aspects of the 
utilization review model and aspects of the demon- 
stration efforts. The model plan which was developed 
consisted of three review components: (1) concurrent 
review (CR), (2) medical care evaluation (MCE) 
studies and (3) profile analysis (PA). 

in concurrent review, admissions are reviewed 
against physician-established criteria for medical 
necessity. Certified admissions are assigned an initial 
number of days according to local diagnosis-specific 
norms of care. Periods of hospitalization extending 
beyond initially certified number of days are reviewed 
to determine whether a continued need for hospitaliza- 
tion exists. Concurrent reviews can either be performed 
by the PSRO (non-delegated) or the hospital 
(delegated). The majority of review (78 percent) is 
currently delegated to hospitals. MCE studies consist 
of in-depth retrospective reviews to determine whether 
certain criteria, thought to assure professionally 
accepted standards of care, were met. Profile analysis 
is statistical analysis of aggregate patient care data, 
conducted retrospectively after patient discharge, 
concerning patterns of health services utilization and 
patterns of care rendered to patients. 

Individual PSROs pass through three stages of 
development and funding. The planning stage 
comprises the development of review plans and 
organizational structure and provides for the enroll- 
ment of physician members. The conditional stage 
may last 48 months and ensures that the PSRO is 
operationally viable before being fully designated. 
Finally, in order to be fully designated, a PSRO must 
operate in a manner consistent with the law, demon- 
strate acceptable financial management, comply 
with Federal reporting requirements and etablish and 
pursue measurable objectives. Within two years after 
full designation a PSRO must initiate review of 
ambulatory services. 

1.3.3 Potential 

The PSRO program is designed to affect hospital 
costs indirectly, by changing the utilization of 
hospitals. However, data indicate that increases in 



inpatient hospital days or admissions are not a major 
factor in the escalating hospital costs. Only six 
percent of the increase in hospital costs between 1970 
and 1978 is due to population growth, while 65 percent 
is due purely to inflation. The remaining 29 percent 
is attributable to increased service intensity defined 
as "greater use and/or changes in the kinds and 
amounts of services provided". ^^ While estimates of 
the relative influence of causal factors vary, it is 
generally agreed that in recent years changes in 
hospital inpatient days of care have represented only 
a secondary influence on the rate of increase in 
hospital expenditures.''^ 

The PSRO program, by itself, was not designed to 
affect hospital resource intensity; rather it was 
intended to affect hospital utilization. Studies indicate 
that inappropriate utilization of hospital facilities does 
exist. Estimates of the extent of inappropriate hospital 
days range from 10 to 35 percent.'* These studies 
indicate that there is considerable potential for 
reducing current levels of hospital utilization, although 
such reductions probably would not result in major 
reductions in reimbursements for hospital care in the 
short run. 

As previously mentioned, PSRO is one of a number 
of governmental regulatory responses to increased 
hospital costs. This report focuses on the PSRO 
program and does not evaluate comprehensively the 
relative impacts of other regulatory mechanisms. 
Given the variety of factors which influence health 
care costs, it is unreasonable to expect any one 
limited program to accomplish hospital cost contain- 
ment alone. This evaluation does not address PSRO 
impact in terms of overall hospital costs, but focuses 
on cost components over which PSROs are to exercise 
influence; namely, the utilization of short-stay hospital 
services and related expenditures associated with 
Medicare and Medicaid beneficiaries.^'' 



' U.S. Congress, The Social Security Act, Title XI, Section 
1151. 



"R. M. Gibson, "National Health Expenditures, 1978," Health 
Care Financing Review (Summer 1979), p. 4; also see 
J. A. McMahon, and D. E. Drake, 'The American Hospital 
Association Perspective," in M. Zubkoff, I. Raskin and 
R. S. Hanft (Eds.), Hospital Cost Containment, (New York: 
Prodist, 1978), pp. 76-102. 

'■ U.S. Congressional Budget Office Expenditures for Health 
Care, Federal Programs and Their Effects (August 1977), 
p. 28; Council on Wage and Price Stability, "Labor- 
Management Innovations in Controlling Cost, Part III", 
Federal Register, (September 17, 1976), pp. 40309 and 
40311; Council on Wage and Price Stability, Executive 
Office of the President, "The Rise of Hospital Cost," 
(January 1977), p. iii. 

'' P. Gertman, "Appropriateness Evaluation Protocol: Develop- 
ment and Methodologic Testing of A New Technique for 
Studying Inappropriate Hospital Utilization," Report 
submitted under Contract #SSA-600-75-0209. 

"The availability of uniform information on Medicare 
beneficiaries permits the most extensive analysis in this 
report. Preliminary analyses of Medicaid data are also 
included. 



1.4 Evaluation Mandate 

The original mandate for ttie evaluation of the PSRO 
program called for the National Professional Standards 
Review Council (NPSRC) ^° to report to the Congress 
and the Secretary, not less than annually, findings 
relative to the PSRO program's effectiveness, accom- 
plishments, and relative regional performance.^! The 
Medicare-Medicaid Anti-Fraud and Abuse Amendments 
of 1977 (P.L. 95-142) ^^ deleted the NPSRC requirement 
to submit annual reports and in its place required the 
Secretary of DHEW to submit to Congress annually a 
full and complete report on the PSRO program's 
status, administration, impact, and cost, with recom- 
mendations for legislative changes. The statute specifi- 
cally called for analysis of patterns of utilization rates 
and changes in medical practices attributable to 
activities of PSRO. The NPSRC, however, is still 
required to perform studies and investigations and 
authorized to make recommendations to the Secretary 
and to the Congress. This report attempts to meet, in 
part, the needs of both the Secretary and the NPSRC. 

1 .5 Evaluation Content and Comparison 
with Prior Evaluations 

The components of the 1979 evaluation are sum- 
marized in Table 1. Each component has its own 
specific hypotheses data bases, and analyses which 
are discussed m detail in the following sections. 

The 1979 PSRO evaluation introduces refinements 
and additional analyses relative to the two previous 
PSRO evaluations. The 1979 Medicare Impact Study 
employs essentially the same model used previously, 
but emphasizes regional analyses. No individual PSRO 
results are presented this year because last year's 
disaggregate analysis did not sufficiently relate 
individual PSRO performance to variation in individual 
PSRO management/administrative capabilities. 
Hospitals included in the Medicare Impact Study data 
base have been revised this year to reflect concerns 
raised by GAO.^^ An additional year of PSRO experi- 
ence, of course, lends a longitudinal perspective 
not previously available. Diagnosis and procedure- 
specific utilization analyses are more complete this 
year and for the first time indicate varying impact 
across selected diagnoses and procedures. 

A Medicaid data section has been included which 
provides insight into the estimation of Medicaid use 
rates; Medicaid rate trends are presented for four 



^A national council of physicians mandated by the original 

PSRO statute to guide the development of the PSRO 

program. 
'^ U.S. Congress, The Social Security Act, Title XI, Section 

1163(f). 
^yj.S. Congress, P.L. 95-142, Section 5 (k). 
" U.S. General Accounting Office, Problems With Evaluating 

The Cost Effectiveness Of Professional Standards Review 

Organizations, 1979. 



PSRO case study areas. The Medicaid rates are based 
on PSRO PHDDS data, as is an analysis of changes 
in case-mix severity and case-mix adjusted lengths 
of stay. The Concurrent Review Benefit-Cost Analysis 
this year develops benefit components In a more 
systematic fashion and discusses the indirect impact 
of the PSRO program on non-Federal patients as well 
as providing a benefit-cost ratio related directly to 
Medicare program expenditures. 

The Quality Section of this year's report is consid- 
erably expanded. The variation rate study is now a 
finished work rather than a pilot study, and an MCE 
benefit-cost methodology has been developed and 
tested. A discussion of profile analysis augments the 
discussion of MCEs. A final addition to this year's 
report is an organizational analysis of the PSRO 
program which provides insight into the structure of 
the program and how the program fits into the broader 
context of the health care sector, 

1.6 Findings 

1.6.1 Medicare Impact Study (100% Medicare 
Claims File Study) 

The purpose of the Medicare Impact Study is to 
assess what impact, if any, the PSRO program has had 
in reducing inpatient hospital utilization by Medicare 
beneficiaries. The study is based on a comparative 
analysis wherein all 108 areas with active PSROs 
(defined as PSROs with at least one hospital under 
review as of July, 1978) are compared to all 81 areas 
without active PSROs for the period 1974 to 1978. 
The actual days of care per thousand aged Medicare 
beneficiaries are presented in Figure 1 for the active 
and inactive PSRO areas for the entire United States 
and in Figure 2 for each of the four major Census 
regions. 

PSRO program effects are estimated by comparing 
the relative rates of change in hospital utilization 
between active and inactive areas. If hospital utilization 
rates are observed to fall more rapidly (or rise less 
rapidly) in active than in inactive areas, the PSRO 
program is credited with utilization impact. The 
primary measure of utilization impact is the estimated 
reduction in total days of care per 1000 aged Medicare 
beneficiaries. 

Forced order multiple regression is used to test for 
PSRO impact. The PSRO variable is entered into the 
equation after entering other demographic and health 
care supply variables. Thus, the test of significance 
for PSRO impact is performed after controlling for the 
effects of those other variables which could affect 
hospital utilization. As in last year's study, this year's 
analysis of the 100 Percent Medicare Claims File 
indicates that the PSRO program had a small impact 
on hospital days of care per 1000 aged Medicare 
beneficiaries. While the effect that the PSRO is 
estimated to have had in 1978 Is In the expected 
direction, it does not meet standard levels of 
statistical significance. 



Figure 1 

Total Days of Short-Stay Hospital Care 
Per 1,000 Aged Medicare Beneficiaries, 
By Active and Inactive PSROs, for years 
1974 through 1978: Total United States 



3750 



)700 



CO 

CO 

O 



3650 



3600 




3550 



1974 



1975 



1976 



— 1 

1977 



1978 



Legend ; 



Inactive 
Active 



Figure 2 

Total Days of Short-Stay Hospital Care Per 1,000 Aged Medicare Beneficiaries, 
By Active and Inactive PSROs, for years 1974 through 1978 



\ 




/ 
/ 
J 
1 

/ 
/ 




/ 




/ 




\ \ 


J3 


N. \ 






3 
O 


\ \ 


cn 


\ \ 




\^ \ 


o 


\ \ 






/ 






/ 






1 






1 






1 







/ ; 

/ / 


T-\ 


/ 


U / 


/ 


^ / 




g / 


/ 


u / 




•^ \ 


■ 


^ \ 


\ 


1-1 \ 




o \ 


\ 


z \ 






\ 


y~^ \ 


\ 


to \ 


'-^ \ 


\ 




/ 




/ 
/ 




/ 



<: 5 



TABLE 1 
Components of the 1979 Evaluation Study 



Study 



Purpose or Intent 



Data Source 



Year of 
Data 



PSRO Coverage 



1. 1979 Utilization Studies 



PSRO Impact on 
Medicare Hospital 
Utilization Study 



Evaluative: PSRO impact 
on days of care/1000 
Medicare beneficiaries 

(DOC); discharges/1000 
Medicare beneficiaries 
(DISC); average length 
of stay (ALOS) 



100% Medicare Claims File 1974-1978 

Query File 1974-1978 

Master Facility Inventory 1973-1976 

Area Resource File 1974-1976 

Provider of Service File 1978 

PSRO Quarterly Implementa- 1976-1978 

tion Reports 

Medicare Master Enrollment 1974-1978 

File 



All active and 
inactive PSR areas 
(108 active, 81 
inactive) 
All areas 

All areas 

All areas 
All areas 
All areas 

All areas 



PSRO Impact on 
Specific Diagnosis 
and Procedure 
Categories Study 



Evaluative: PSRO impact 
on days of care/1000 
Medicare beneficiaries 
(DOC); discharges/1000 
Medicare beneficiaries 
(DISC); average length 
of stay (ALOS); for one 
diagnosis (myocardial 
infarction), and four 
procedures (cholecys- 
tectomy, hysterectomy, 
lens extraction, mas- 
tectomy) 



20% Medicare Discharge File 1973-1977 



Master Facility Inventory 1973-1976 

Area Resource File 1974-1976 

Medicare Master Enrollment 1973-1977 
File 



All active and 
inactive PSRO areas, 
as of 1977 (95 active, 
94 inactive) 

All areas 
All areas 
All areas 



Medicaid Hospital 
Utilization Rate 
Analysis 



Developmental: Develop 
measures of days of 
care/1000 Medicaid bene- 
ficiaries; discharges/ 
1000 Medicaid benefi- 
caries; average length of 
stay 



PSRO Hospital Discharge 1976-1978 

Data Set (PHDDS) 

Medicaid Eligibility 1976-1978 



Four active PSROs, 
no inactive areas. 
Four active PSROs, 
no inactive areas. 



Case-Mix Adjusted 
Length of Stay 
Analysis 



Developmental: Changes 
in case-mix adjusted 
lengths of stay; changes 
in case-mix severity 



PSRO Hospital Discharge 
Data Set (PHDDS) 



1977-1978 



12 active PSRO 
areas. No inactive 
areas. 





2. Benefit-Cost Analysis of Concurrent 


Review 






Benefit Analysis 


Evaluative 


Audited Medicare Provider 
Cost Reports 


FY 1978 


3294 hospitals 


Concurrent Review 
Expenditure Analysis 


Evaluative 


PSRO Management 
Information System 




Calendar 
year 1978 


109 PSROs 


3. Program Costs 


Unit Cost Variation 


Descriptive 


PSRO Management 
Information System 




FY 1978 


119 PSROs 


4. Medical Care Evaluation Studies 


Development and 
Current Status 
of MCE Activity 


Descriptive of MCE 
studies conducted 


PMIS routine reports 




1975-1978 


141 active PSROs 


Variation Rate 
Analysis 


Assess changes in com- 
pliance to criteria of care 


Survey of PSROs and 
hospitals 




1976-1979 


62 active PSROs 


Benefits and Costs 
of MCEs 


Assess health and financial 
consequences of MCEs 


Survey of PSROs 




1979 


8 active PSROs 



Table continued on next page 



TABLE 1 (Continued) 



Study 




Purpose or Intent 




Data Source 


Year of 
Data 


PSRO Coverage 


5. Profile Analysis 


Current Status 
Profile Activity 


of 


Descriptive of profile 
studies conducted 




PMIS routine reports 
Case studies 


1979 
1979 


123 active PSROs 
7 active PSROs 






6. 


Organizational Characteristics 






Organizational 




Developmental 




Research literature 
PSRO program materials 
Site visits 


1979 


3 active PSROs 








7. 


1979 Program Status 






Program Status 




Descriptive 




Interviews with HSBQ Staff 
Review of PSRO grant applica- 
tions and contract proposals 


1979 


192 PSROs, active 
and inactive 


Project Officer 
PSRO Performance 
Ratings 


Descriptive 




Assessments by HSBQ Project 
Officers 


1978 


109 active PSROs 



However, there is a statistically significant inter- 
action between PSRO and Census region variables. 
The Census region variables serve as proxy measures 
for the blend of socio-economic, cultural, and geo- 
graphic differences that exist over and above the 
specific poverty, population density, and medical care 
supply variables included as adjustment variables in 
the regression analysis. This indicates there is PSRO 
impact, but that it differs by region. These regional 
variations were also reflected in the analysis con- 
ducted last year. Given the significance of the inter- 
action term, it is appropriate to test for a PSRO effect 
in each of the four Census regions separately, the 
results of which are presented in Table 2. This table 
is discussed in detail in Section 2.2. 

The results of the regional analyses show that PSRO 
areas in the Northeast and North Central regions have 
a statistically significant impact in reducing the 



l\/ledicare days of care rate. The PSRO impact in the 
West region is in the direction of reducing days of care 
but does not attain statistical significance, in the 
South region the analysis indicates that PSRO activity 
apparently contributes to a significant increase in the 
days of care rate. The data suggest small relative 
declines in active PSROs, and rather striking regional 
differences in PSRO impact. 

The Medicare Impact Study utilizes the regression 
results to estimate the number of Medicare "days 
saved" due to PSRO review activity in 1978. This 
measure of days saved is then transformed into a 
dollar value savings in the Benefit-Cost chapter 
(Section 3). Days saved are calculated for each 
Census region as shown in Table 2. The estimated 
number of Medicare days saved for 1978 is 948,430 
days, representing 1.7 percent of Medicare days of 
hospital care in active PSRO areas. 



TABLE 2 
Estimated PSRO Impact on Days of Care per 1000 Aged Medicare Beneficiaries, by Region 





(1) 


(2) 


(3) 


(4) 


(5) 


(6) 




Estimated 


Estimated 


PSRO Impact 








Region 


Days of 


Days of 


on Days of 


PSRO Percent 


Medicare 


Estimated 




Care /1 000 


Care/1000 


Care/1000 


Impact 


Aged 


Days Saved 




Without PSRO With PSRO 


(2)-(1) 


(3)^(1) 


Enrollment ^ 


(3) X (5)^1 000* 


Northeast 


4240 


4037 


—203 


-4.8% 


4,728,898 


—959,924 


North Central 


4106 


4020 


—87 


-2.1% 


3,783,674 


—328,612 


South 


3691 


3826 


+135 


+3.7% 


3,383,325 


+456,039 


West 3 


2787 


2748 


—39 


—1.4% 


2,958,974 


—115,933 


Nation 


3791 


3728 


—64 


1.7% 


14,854,871 


—948,430 



^ Enrollment is for active PSRO areas only. 

" Column (6) does not equal (3) x (5) h- 1000 exactly due to rounding of column (3). 

^The estimate of PSRO impact in the West is based on an insignificant regression coefficient. 



1.6.2 Diagnosis- and Procedure-Specific Impact Study 
(20 Percent l\1edicare Discharge File) 

The second analysis of Medicare hospital utilization 
rates tests hypotheses about PSRO impact on specific 
medical diagnoses and surgical procedures. This 
study presents a priori hypotheses that PSRO inter- 
vention is or is not likely to reduce the specific 
utilization rates of selected diagnoses and procedures. 
The methodology used in the 1978 PSRO Evaluation 
analysis of Medicare utilization rates is applied here 
to specific diagnoses and procedures. 

Diagnoses and procedures are available from a 20 
percent sample of Medicare hospital claims. Since the 
20 percent file is current only through 1977, the 
Diagnosis and Procedure-Specific Impact Study is 
not an evaluation of the most recent PSRO experience, 
but rather an elaboration of the 1978 Evaluation 
finding that PSROs reduced overall days of hospital 
care per 1000 aged Medicare beneficiaries for the 
period 1974 to 1977. The current analysis addresses 
the question of whether the decrease was concentrated 
in specific procedures and diagnoses or represented 
a more general reduction across all diagnoses and 
procedures. 

As a result of a review of the medical literature, one 
diagnosis and four procedures were selected for which 
the PSRO progam would be expected to have an 
impact in shifting physician practice toward less 
expensive alternative modes of care. The diagnosis 
is acute myocardial infarction, where it was hypothe- 
sized that PSROs reduce the average length of stay; 
the procedures are cholecystectomy, cataract removal, 
and hysterectomy, where it was hypothesized that 
PSROs reduce the discharge rate, and breast cancer, 
where it was hypothesized that PSROs reduce the 
proportion of radical mastectomies to total mastec- 
tomies. For each diagnosis or procedure, descriptive 
statistics are used to assess the changes in the 
utilization measure of interest for active and inactive 
PSROs from 1973 to 1977. This is followed by a forced 
order multiple regression to test the statistical 
significance of the impact of active PSROs relative to 
inactive PSROs. 

The results indicate statistically significant PSRO 
effects for four of the five diagnoses and procedures 
tested. As hypothesized, active PSROs did yield a 
greater decline in average length of stay for acute 
myocardial infarction, and in discharge rate for 
cholecystectomy. The cataract surgery discharge rate 
increased for both active and inactive PSROs, but the 
rate of increase was significantly lower for active 
PSROs, a finding consistent with the hypothesis. 
Finally, there is a statistically significant PSRO effect 
on the use of radical mastectomies as a percentage 
of total mastectomies from 1973 to 1977; while the 
younger PSROs show an increase in discharge rate, 
the older ones show a decrease, resulting in an overall 
favorable impact. Only in the case of hysterectomy is 
there no significant PSRO effect. 



1.6.3 Medicaid Hospital Utilization Rate Analysis 

While prior PSRO evaluations focused on the impact 
of PSRO review activities upon Medicare hospital 
utilization, there has never been a national analysis of 
the impact of PSRO on Medicaid hospital utilization. 
This lack of analysis is due to the fact that there have 
been no national statistics for Medicaid hospital 
utilization or eligibility. The Medicaid Hospital Utiliza- 
tion Rate Analysis explores the potential use of two 
existing data sources, the PSRO Hospital Discharge 
Data Set (PHDDS) and Medicaid State Agency eligi- 
bility data, to develop rate analyses and applies the 
methodology to four case study PSROs. 

The study attempts to develop standardized meas- 
ures of utilization, e.g., rate variables, as a basis for 
studying differences in utilization among PSRO areas 
over time. Medicaid rate studies have been limited 
by both the availability of data and the inconsistency 
of the data among the States. This is a result of the 
fact that the Medicaid program is a collection of 
programs for medical assistance that are administered 
by individual States. Each State program contains 
nuances with respect to eligibility, coverage, benefit 
amounts, administration, and statistical reporting that 
confound analytical as well as evaluative efforts. Given 
these difficulties, the objectives of this study were to: 

• Explore the use of PHDDS utilization data as 
numerators for Medicaid rate analyses; 

• Explore the use of State Medicaid Agency eligi- 
bility data as denominators for Medicaid rate 
analyses; 

• Develop appropriate methodology for the linking 
of PHDDS data to State Medicaid Agency eligibility 
data in rate construction; 

• Calculate Medicaid rates and analyze trends for 
each of four PSRO case studies; and 

• Recommend further developmental work related 
to data acquisition and accompanying analytical 
framework. 

The study uses the PHDDS as numerator utilization 
data because the PHDDS has the advantages of being 
a Federally administered system that is generally 
available at low cost. PHDDS data are limited in their 
use for evaluation purposes, however, in that they are 
available only for active PSRO areas. 

Days of care, discharge rates per 1,000 eligibles, 
and average length of stay were studied within the 
four distinct PSRO case studies. Hypotheses relating 
to change in these measures were treated for all dis- 
charges, twenty-two selected diagnosis groups, and 
four surgical status groups. 

Counts of total discharges, total days of care, and 
eligibles tended to fluctuate from quarter to quarter 
in all four case study areas. Utilization rates tended 
to fluctuate as well. Days of care rate trends were 
mixed — two PSROs exhibited declining patterns while 
two PSROs showed increasing patterns. None ofthe 
days of care rate changes was statistically significant. 
As observed in the prior PSRO evaluations, increases 



(decreases) in days of care rates were associated with 
increases (decreases) in discharge rates. Analyses for 
specific diagnostic groups exhibited a similar pattern. 
Both discharge and days of care rates for several 
mentally related disorders showed increases for all 
four case study areas. 

Utilization rates were not compared among PSROs. 
It was not possible to adjust for outmigration and for 
variations in the mix of Medicaid program eligibility 
across PSRO areas. 

The recommendations of this section represent 
alternatives along a continuum of actions designed to 
provide greater flexibility in future analytical design. 
Since PHDDS data contain no pre-PSRO data or non- 
PSRO data to serve as control areas, the extensions 
of PHDDS analysis recommended in this section 
provide only limited enhancement to evaluative power. 
Still the recommended extensions would be useful 
since PHDDS data contain relatively consistent coding 
and are under direct administrative control of the 
Federal government. PHDDS analyses should therefore 
continue, but the more desirable long-range choice 
for studying Medicaid utilization rates is to implement 
the study's recommendations for construction and 
analysis of State Medicaid claims files. 

1.6.4 Case-Mix Adjusted Length of Stay Analysis 

This study is a developmental analysis in which the 
average length of stay (ALOS) characteristics of 
various PSROs were adjusted for variations in case- 
mix. The data base used was the PSRO Hospital 
Discharge Data Set (PHDDS); the specific case-mix 
adjustments were based on patient age plus two 
additional factors reflective of case-mix differences: 
single vs. multiple diagnoses and surgical vs. non- 
surgical stay. Since older patients, patients with 
multiple diagnoses and patients with surgery tend to 
have longer lengths of stay, the case-mix adjustments 
would tend to clarify changes in the unadjusted ALOS 
as the proportion of these factors changes. 

Two indices are developed. The first represents a 
measure of ALOS which adjusts for variations in 
patient case-mix; the second is an index which reflects 
variations in case-mix proportions. Both measures are 
applicable at the individual PSRO level; both may be 
used either longitudinally, to study trends over time, 
or cross-sectional ly, to compare the performance of 
PSROs and groups of PSROs. Comparisons can also 
be made between Medicare and Medicaid 
beneficiaries. 

These measures are applied to data covering dis- 
charges reported by twelve PSROs during 1977 and 
1978. The PSROs selected for the study are those 
which had submitted PHDDS data of acceptable 
quality to HSQB during the full two-year period. Both 
Medicare and Medicaid beneficiaries are included in 
the analysis. 

Two sets of results are produced. The first set, 
which includes longitudinal changes over time, shows 
that almost all of the PSROs in the study experienced 



shifts in case-mix toward those cases which normally 
have longer stays, and that as a consequence their 
adjusted ALOS changes were more favorable 
(decreased by a greater percentage) than the 
unadjusted figures would indicate. These findings, 
which are statistically significant, are an encouraging 
sign; they imply that the study PSROs tended to 
accomplish two of the functions expected of effective 
PSROs: a shift toward a more complex case-mix and 
a decrease in the average length of stay for patients 
of a given type. The relevant figures are presented in 
Table 3 for the total patient population. When the 
Medicare and Medicaid populations were studied 
separately, the shift toward cases which tend toward 
longer lengths of stay was less pronounced for the 
Medicaid population than for Medicare. 

The second analysis in this study is cross-sectional 
and shows that ALOS differences among PSROs can 
be factored into two components: one component 
attributable to differences in case-mix, the other to 
differences in true ALOS behavior. 

The techniques developed in the study are in a 
relatively early stage of development, but initial results 
taken from the twelve case study PSROs suggest the 
potential usefulness of these techniques for both 
longitudinal and cross-sectional analyses. 

TABLE 3 
Unadjusted vs. Adjusted Indices, 12 PSROs 

Unadjusted Adjusted Indices' 



1st quarter, 1977 
2nd quarter, 1977 
3rd quarter, 1977 
4th quarter, 1977 
1st quarter, 1978 
2nd quarter, 1978 
3rd quarter, 1978 
4th quarter, 1978 

* 1st quarter 1977 = 100 

1.6.5 PSRO Benefit-Cost Analysis 

The benefit-cost methodology estimates the dollar 
value of the PSRO-related reduction in days of care for 
Medicare patients 2* and relates these dollar benefits 
to concurrent review costs for Medicare hospital 
admissions. The resulting benefit-cost ratios indicate 
the number of dollars saved (on the average) for each 
dollar expended for Medicare concurrent review. 

Since the perspective of the benefit-cost analysis 
is the Medicare program, benefits are limited to 
changes in reimbursements associated with Medicare 
concurrent review. Benefits resulting from review of 
Medicaid and Title V patients are excluded, as are the 
benefits of other forms of Medicare review such as 



<\LOS* 


CM 


CM /ALOS 


100.00 


100.0 


100.0 


97.2 


102.1 


95.1 


95.6 


102.0 


93.8 


97.8 


103.1 


95.2 


100.7 


102.4 


98.2 


95.8 


103.4 


92.8 


97.5 


103.5 


94.3 


97.1 


103.1 


94.1 



-*As derived by the Medicare Impact Study (Section 1.6.1). 



10 



TABLE 4 



Benefit-Cost Ratios for Three Alternative Assumptions Concerning 
Ambulatory Substitution Effects 



Census Region 



=0.3 



S,=0.5 



Sa=0.7 



Northeast 
North Central 
South 
West 
Nation 



4.58 
2.13 
—2.28 
0.87 
1.504 



3.88 
1.79 
-1.91 
0.72 
1.269 



3.17 
1.44 
-1.54 
0.57 
1.035 



areawide or MCE review. The data pertain to CY 1978 
when the PSRO program was not fully implemented. 
The research results of this evaluation apply most 
directly to that time period. 

The benefits are estimated by using the reduction 
in Medicare days from the Medicare Impact Study, 
hospital-specific cost data for CY 1978, and the 
Medicare hospital reimbursement formula to estimate 
the change in reimbursements associated with "saved" 
Medicare days. There are four components to these 
potential savings, corresponding. to the four types of 
reimbursement adjustments that occur within the 
reimbursement formula. They include a variable cost 
effect for both routine and ancillary services and a 
reimbursement formula realloction effect for both 
routine and ancillary services. 

Potential gross savings are calculated at the hospital 
level and then aggregated to the Census region. 
Because the PSRO-regional interaction variable in the 
Medicare Impact Study is significant, separate benefit- 
cost ratios are derived by Census region and then 
aggregated across Census regions. 

After potential gross savings are calculated, they 
are reduced by two factors which account for substi- 
tution effects. The specific substitution effects treated 
are: (1) the substitution of long-term care (LTC) days 
for inpatient short-stay hospital days and (2) the sub- 
stitution of outpatient ambulatory care for inpatient 
ancillary services. 

For purposes of the benefit-cost calculations, it is 
assumed that for every 100 days saved in the hospital. 
Medicare will pay for an additional 15 days of LTC. 
The degree to which reduction in inpatient hospital 
days ultimately leads to increases in ambulatory care 
has been a point of debate. Since the value used for 
the ambulatory care substitution effect is critical to the 
benefit calculations, ambulatory care substitution 
effects of 30, 50, and 70 percent are used for purposes 
of sensitivity analysis. 

Table 4 displays calculated benefit-cost ratios. A 
separate set of ratios is produced for each of three 
assumed ambulatory substitution rates (0.3, 0.5, and 
0.7). As shown in the table, the 0.5 substitution rate 
used last year, which produced a 1.1 benefit-cost ratio 
for 1977, gives a national ratio of 1.269 for 1978. 

With an ambulatory substitution rate of 0.3, the 
benefit-cost ratio reaches 1.504 and at 0.7 it is 1.035. 



In all three cases, the West has a benefit-cost ratio 
less than one, the positive savings in the North Central 
are slightly more than offset by the negative savings 
in the South, and the Northeast exhibits a benefit-cost 
ratio ranging from 3.17 to 4.58. 

While these benefit-cost ratios apply to the Medicare 
program, from the societal perspective one should 
take into account the transfer of costs to the non- 
Medicare population resulting from the reimbursement 
formula. When this issue is examined in Section 3, it 
is found that up to half of the potential unadjusted 
savings in reimbursements may represent fixed costs 
which are absorbed by other payors. 

Two factors discussed in Section 3, but not included 
in the calculations, are the PSRO-induced inpatient 
reduction in Part B (Medicare Physician Services) 
reimbursements and the potential impact of PSRO on 
private patient hospital utilization. Supplementary 
studies are presently being developed in each of 
these areas. 

1.6.6 Overview of PSRO Review Costs 

Data on review costs for individual PSROs are 
tabulated from the Unit Cost Report for CY 1978 pre- 
pared by HSQB. The objective is to examine the 
magnitude of, and variation in, different review cost 
components and to test the hypothesis that small or 
newly implemented PSROs experience higher costs 
than larger and more mature PSROs. 

As shown in Table 5, median concurrent and MCE 
review cost per discharge declined between 1977 and 
1978. This decline, however, is more than offset by an 
increase in areawide review costs. 2"' The result is a net 
increase in total hospital median review costs. Federal 
management and support costs also increased slightly. 
The range of costs for most cost components was 
narrower in 1978 than 1977. Total hospital review costs 
per discharge range from $6.92 to $30.74 with a 
median of $1 2.91 . The $1 2.91 figure is considerably 
greater than the $8.70 per discharge target set for the 
end of FY 1979. 



'Areawide review costs include the costs of physician and 
patient profiles, the cost of computer support, and the 
cost of monitoring delegated review functions. 



11 



TABLE 5 

National Summary of PSRO Review Cost Components: A Comparison of Results 
from the 1978 and 1979 Evaluations 



Hospital-Based Review Cost 



Year 



Concurrent 
Review 



i\/ICE 
Review 



Areawide 
Review 



Federal 
Total Hospital Management 
Review Costs ^o and 

Support Costs 



Median PSRO 
Review Cost 
Per Discharge: 
Cost Range: 



1977 
1978 
1977 
1978 



$8.76 
8.57 
3.25-18.09 
2.75-21.68 



1.40 
1.19 
.08-9.19 
0-4.01 



2.21 

3.24 
.08-59.07 
.01-14.13 



12.31 

12.91 
6.52-74.90 
6.92-30.74 



5.07 
5.33 
1.11-81.44 
1.13-70.60 



Since ttiere is no information on the number of 
reviews conducted, there is no way to calculate the 
cosf per review. However, the costs per discharge do 
not differ substantially between the small and large 
PSROs. 

1.6.7 Quality of Care Studies 

1.6.7.1 MCE VARIATION RATE STUDY 

The impact of medical care evaluation (MCE) 
studies on the quality of care provided in PSRO 
hospitals is examined by collecting data from a 
representative sample of MCE audits and reaudits. 
The measure of quality is the variation rate, defined as 
the proportion of patient records audited to a particular 
criterion of care which do not meet the standard set 
for the criterion. In previous PSRO evaluations, the 
variation rate was validated as a measure of change 
in quality of care but this is the first time it has been 
applied to a representative sample of MCEs. 

The MCEs studied include 710 audit and reaudit 
combinations in six medical diagnoses and six surgical 
procedures. Criteria of care are sampled from each 
of the MCEs, and variation rates are calculated before 
(at audit) and after (at reaudit) actions presumably 
were taken to correct problems in care. The principal 
question was whether the variation rates were reduced 
significantly from audit to reaudit. 

The study results show a statistically significant 
trend: variation rates at reaudit tend to be lower than 
those at audit, indicating measured improvements in 
quality of care. This is especially true when criteria 
whose initial variation rate was 10% or more were 
examined separately. These are the criteria which 
reflect more serious problems in care, and which are 
more likely to be acted upon by medical audit com- 
mittees. For these criteria, the improvement between 
audit and reaudit was pronounced. 



'Total hospital review costs include the costs of concurrent, 
MCE, and areawide review. However, the median statistics 
are not additive. 



1.6.7.2 BENEFITS AND COSTS OF MCEs 

The second quality-of-care study is more exploratory. 
Variation rate changes in a sample of MCEs are used 
as a basis for estimating health benefits resulting from 
MCEs. The benefits are then related to two kinds of 
costs: the costs of conducting the MCEs, and the 
costs of the changes in care brought about as a result. 
The findings indicate that even small improvements 
in variation rates can have major implications for 
patient health status, and that when the monetary 
value of improved longevity and functioning are 
compared to MCE costs, the benefits far outweigh the 
costs. 

1.6.8 Profile Analysis 

Profile analysis, generally the last component of the 
PSRO review system to be implemented, is receiving 
emphasis from the program as a way of increasing the 
efficiency of review. Under a new reporting system, 
123 PSROs indicated some sort of profile activity in 
1979. While PSROs seem to be becoming more 
accomplished in their use of data, profiling has yet to 
achieve a guiding function in all but the most 
advanced PSROs. 

1.6.9 Organization Models and Performance 
Assessments 

Past PSRO evaluations have contained little work 
on the organizational characteristics of PSROs and 
how those characteristics might be expected to and do 
affect PSRO effectiveness. This report contains a 
review of the research literature relevant to the study 
of PSRO organizational effectiveness as well as 
preliminary observations of PSRO performance. PSROs 
are described as formalized, externally authorized, and 
mandated local physician organizations expected to 
function as a regulatory system exercising control 
via performance evaluation tied to financial and 
professional sanctions. In their capacity as locally 
organized physician regulatory systems, the PSROs 
incorporate features of earlier control systems govern- 
ing the delivery of care in hospitals; however, insofar 
as they are more highly formalized, are externally 



12 



validated and mandated and have their decisions 
backed by financial as well as professional sanctions, 
PSROs represent a new form of control over medical 
practice. 

Two important functions of the PSRO should have a 
positive impact on the quality assurance and cost 
reduction activities in hospitals. First, the information 
gathering and processing activities of the PSRO make 
it, at least potentially, a force for change and reform. 
Gradually, data systems are becoming operational. 
The availability of comparable data sets for all 
hospitals will permit the identification of "outliers," 
those hospitals and physicians whose performance is 
substantially at variance from the standards of care 
provided by the average hospitals. Second, PSROs 
can have some impact on the performance of hospitals 
by the types of training and support which they 
provide for the hospital review coordinators and 
physician advisors. Prior to the existence of PSROs, 
utilization review and quality assurance personnel 
functioned within the confines of their own hospitals, 
often limited in the types of data which they could 
collect and lacking information on the performance 
levels achieved in comparable hospitals. Thus, the 
PSRO program has brought more uniformity, and 
possibly more leverage, to the utilization review 
process. 

A number of observers, however, have questioned 
the potential efficacy of the PSROs on theoretical 
grounds." The delegated structure of the review 
activities removes the PSRO from direct contact with 
the physicians and patients for whose care they are 
responsible.28 For financial and philosophical reasons, 
PSROs were initially encouraged by DHEW to delegate 
the actual review process to hospitals, and to monitor 
the performance of those hospital review committees. ^^ 
This "two steps removed" system has been seen as 
being potentially a relatively weak, indirect regulatory 
mechanism. In addition, the feedback mechanisms 
that bring the results of the review efforts to the 
attention of the practicing physicians are also rela- 
tively limited. Most physicians have no regular contacts 
with the review system. Individual physicians are 
contacted only if the physician advisor concurs with 
concerns referred by the review coordinator. The MCE 
studies, which permit in-depth and comparative 
assessments of medical care practice, suffer from 
delays in the gathering and processing of information; 
as a consequence, when feedback to hospitals or 
physicians occurs, it often has not been timely. 

PSROs currently lack control over some of the 
critical aspects of their own functioning. Hospital 
review costs are not covered by appropriated PSRO 
funds, but are reimbursed from the Medicare Trust 
Funds. Each hospital negotiates a financial arrange- 



^' See Sloan and Steinwald, op. cit. 

'^ PSROs are responsible for monitoring delegated hospital 
review effectiveness. Delegated status is intended to be 
given only to those hospitals which have shown, over time, 
evidence of adequate utilization review. 

^P.L. 94-202; H.R. 4000. 



ment with the PSRO. However, although fiscal inter- 
mediaries are required to request PSRO review of and 
comment on all costs claimed by delegated hospitals 
which would result in the hospital being reimbursed 
at a higher unit cost rate than negotiated at the 
beginning of the year, the Medicare fiscal inter- 
mediaries have the final settlement authority. 

The PSROs are armed with a variety of sanctions 
which are mostly negative in nature. No positive 
incentives to reward effective or successful perform- 
ance are built into the operation of the program. The 
typical contacts which occur between practicing 
physicians and PSRO representatives are those 
occasions when a physician's judgment is being 
questioned. Physicians and/or hospitals are not sought 
out in order to be commended for their effective 
efforts. None of the cost savings which are presumably 
produced by actions of the PSRO and its participating 
hospitals are passed on to those hospitals to reinforce 
their continued cooperation. 

In addition to reviews of the literature and observa- 
tions about PSROs, Section 7 addresses the implica- 
tions of using different explanatory models for 
evaluation of PSROs. It is noted that realistic 
performance standards for PSROs must be developed. 

1.6.10 Program Implementation Status 

Section 8, which addresses PSRO program status, 
provides information about the program with respect 
to PSRO costs, management objectives, and review 
activities performed. The sources of information in- 
clude staff of the HCFA Health Standards and Quality 
Bureau (HSQB), PSRO transmittals and other docu- 
ments, and PSRO grant applications and contract 
proposals. 

In fiscal year 1979, the PSRO program operated on 
a budget of $149,866,000. Of the 195 designated PSRO 
areas, three PSROs are currently at the planning stage, 
186 are conditional while the remaining six are still 
unfunded. PSROs are currently implementing review on 
Federal patients in 82 percent of the nation's 5,524 
short-stay hospitals; 49 PSROs are also performing re- 
view on non-Federal patients. One hundred twenty 
three PSROs are submitting PHDDS data to report 
their activities. 

In responding to the general management initiatives 
instituted by HSQB in 1978, 106 PSROs in all ten 
DHEW regions submitted impact objectives in 1979, but 
only a few PSROs have submitted self-assessment 
impact statements to HSQB. 

A review of the 1979 PSRO grant and contract files 
provided information on 192 PSROs. From this source 
it was determined that, for hospital review implemen- 
tation, 69 PSROs, as of the end of 1978, had modified 
their concurrent review methodologies with "focusing," 
a method of reducing the number of reviews, and 
thereby reducing the costs of review. Profile analysis 
activities were reported by 123 PSROs; as noted in 
Section 5, a similar number of PSROs (121) reported 
having conducted medical care evaluation studies. In 



13 



addition to implementing hospital reviews, 48 PSROs 
and 35 PSROs are performing binding long-term care 
review and ancillary care review respectively. 

As a result of a $7.5 million budget cut, the FY 1980 
PSRO budget has been reduced to $144,416,000, and 
the total hospital review costs were limited to $85.77 
million. This significantly affects all aspects of the 
PSRO operation. PSROs are expected to: (1) expand 
data system implementation, objective setting, financial 
management, and self assessment ;(2) improve rela- 
tionships with external agencies; and (3) increase im- 
plementation of focused review, profile analysis, surgi- 
cal review, long-term care review, ambulatory review, 
ancillary review and physician services review. Yet, 
with the exception of $1,979,080 funded for special 
initiatives in ancillary care and other review, no addi- 
tional funding is available for program expansion. 

1.7 Conclusions 

In summary, the individual studies described present 
a picture of the PSRO program as a developing form 
of regulatory mechanism which is being established in 
almost all parts of the country for the purpose of con- 
trolling the hospital costs and assuring the quality of 
care associated with Federal health care beneficiaries. 
With respect to the cost control objective, the PSRO 
program has shown the ability to make a modest impact 
on reducing or slowing down increases in Medicare 
beneficiary hospital utilization, with an estimated sav- 
ings in Medicare expenditures $21 million greater than 
the cost to administer the PSRO program. The impact 
on hospital utilization is not uniform, but varies mark- 
edly in different parts of the country. With respect to 
the quality assurance objective, the PSRO program 
appears to be producing increased physician com- 
pliance with quality of care criteria and facilitating 
the implementation of information about efficacious 
medical procedures and treatments. 

This pattern of findings is consistent with the find- 
ings of the 1978 evaluation analyses, and not incon- 
sistent with the 1977 evaluation of the earliest phase of 
the PSRO implementation. It has been shown that im- 
pact does not occur immediately, but increases with 
time, and then seemingly flattens out. The longevity of 
even the oldest PSROs is not really sufficient to de- 
termine the long-range impact potential of the PSRO 
regulation mechanism. 

The existing body of research on the effectiveness 
of the certificate of need procedure, health planning, 
and hospital rate setting is not encouraging. For the 
second year the PSRO analyses indicate that state- 
wide hospital rate setting is associated with increased 
hospital utilization for Medicare beneficiaries. While a 
sophisticated evaluation of the efficacy of statewide 
hospital rate commissions is being planned by HCFA, 
preliminary findings using crude measures cause con- 
cern as to the eventual results. 

There is statistical evidence that the PSRO program 
is having a modest desired effect in both the cost and 
quality aspects of the program. However, the severe 



budgetary constaints on the PSRO program may pre- 
clude the program from ever developing to its fullest 
potential. The PSRO program needs a clear Congres- 
sional endorsement if it is to continue to be effective. 

It is important to stress that the results of the evalua- 
tion apply to the program in its state of implementa- 
tion during CY 1978; these findings may not be legiti- 
mately interpreted to represent the potential of a fully 
implemented program. For example, if one were to 
extend the results of the utilization study, the findings 
for the South would need to be weighted relatively 
greater than those for the other three regions since 
the South is currently the least implemented of the 
four regions. Although the regression results for the 
South indicate that PSRO review has thus far been 
associated with added days of care, a comparison of 
adjustment variables across active and inactive areas 
in the South suggests that the inactive areas are highly 
dissimilar from the actives. Thus it is not clear that 
expansion of the PSRO program in the South will con- 
tinue to produce unfavorable results. 

Another qualification to the extension of the results 
of this evaluation concerns the assumed ratio of vari- 
able to fixed hospital costs. In this study a variable- 
fixed cost ratio of 40:60 was assumed. However, with 
a longer planning horizon hospitals will have greater 
discretion over investments and a smaller percentage 
of their total costs will be fixed. Also, as cost contain- 
ment becomes a key concern of more government 
agencies, there is an increased likelihood that hospital 
capacity will be reduced — i.e., relative to present 
trends, the supply of beds and capital equipment may 
decrease. In that event, 40 percent would be an under- 
estimate of the variable costs for a fully implemented 
program; the 40:60 assumption would therefore under- 
estimate the impact of PSRO review on reductions in 
Medicare reimbursements. 

A further argument against the extension of the 
present research results stems from the expectation 
that the PSRO review budget will not increase in the 
near future. While small increases in the cost of review 
are expected, as more areas are implemented and 
existing PSROs improve, the benefits are expected to 
increase. Hence, on balance, one may expect the bene- 
fit-cost ratios to increase. 

A final caveat concerns the self-selected nature of 
the active PSROs versus those which are relatively 
inactive. Like most other Federally legislated programs, 
the PSRO program was mandated to be implemented 
in all parts of the country as soon as possible. In order 
to implement quickly, the Department designated 
volunteer applicants as PSROs once they were judged 
to be qualified. A large body of literature discusses the 
potential bias of "volunteers;" not only may the first 
volunteers into a new program be atypical, but so may 
the "holdouts." This element of self-selection of "ac- 
tive" PSROs thus produces an overarching limitation on 
all evaluations undertaken: for this reason as well as 
for those cited earlier one cannot reliably extrapolate 
from the experience of the partially implemented pro- 
gram to that of a fully implemented program. 



14 



The fact that the PSRO program was not imple- to which one can assume that currently inactive 

mented using standard experimental design method- PSROs will experience the same patterns of impact as 

ology does not, however, invalidate the results of those which are active, 
the PSRO experience to date; it only limits the extent 



15 



2. Utilization Impact Studies 
2.1 Overview of Utilization Studies 

2.1.1 Objectives 

PSROs undertake utilization review to accomplish 
two interrelated goals: first, to assure that the quality 
of medical care provided to Federal beneficiaries 
(Medicare, Medicaid, and Maternal and Child Health) 
is appropriate, and second, to control excess utiliza- 
tion of health services by these beneficiaries. Due to 
the continuing rapid growth in health care costs, and 
the consequent increase in the share of the Federal 
budget allocated to medical services in recent years, 
relatively greater Congressional emphasis has been 
placed on the potential effectiveness of PSROs in 
controlling utilization than upon the corollary goal of 
maintaining quality. To date, PSRO review has focused 
on inpatient hospital utilization. A small number of 
PSROs are beginning to do long term care review and 
an even smaller number have started ambulatory care 
review. This emphasis on hospital utilization review 
seems appropriate since hospitals consume 40 per- 
cent of the health care dollar, more than any other 
sector of the health care marketplace. 

The nation's PSROs have developed two mech- 
anisms of concurrent review to control utilization: 
admission review and continued stay review. Admission 
review is undertaken to assure the necessity and ap- 
propriateness of hospital admissions while continued 
stay review is directed toward preventing excessive 
lengths of stay. Both types of review have support 
from an extensive body of literature which in- 
dicates that in many cases care is more appropriately 
rendered in outpatient settings and that many patients 
continue to stay in hospitals when they no longer need 
to be there. Together, admission review and continued 
stay review are thought to provide a potential for in- 
fluencing hospital utilization. 

The primary objective of the studies reported in this 
section is to determine whether PSRO utilization review 
activities have indeed resulted in a decrease in 
hospital utilization. 

The hospital use measures (outcome measures) em- 
ployed in this chapter are days of care, admissions 
(discharges) and average length of stay. Days of care is 
considered the primary outcome measure since it is 
most directly related to hospital expenditure levels. 
While discharges and average length of stay are viewed 
primarily as components of days of care, and as such 
receive secondary consideration, they each have dif- 
ferent programmatic implications and are thus of in- 
terest to program managers. 

Days of care and discharges are estimated in rate 
form (e.g., days of care per 1000 Medicare enrollees) 
wherever possible. Rates are superior to simple, 
"non-standardized" counts in that they adjust for 
changes in the population at risk and thus give a 
clearer indication of the relationship between PSRO 
activity and hospital utilization. 



When reductions in days of care rates are attributed 
to the PSRO program, it is assumed in these studies 
that such reductions represent the elimination of un- 
necessary care. Accordingly, reductions in hospital 
use do not necessarily imply reductions in the quality 
of care received. This assumption concerning quality 
of care seems reasonable in that the range of inap- 
propriate utilization typically reported is between 9 
and 20 percent of total days,i"^ while the extent of 
PSRO impact on days of care reported below is on 
the order of 1 to 5 percent. It appears that there is 
sufficient "slack" in the system that relatively large 
reductions in use could occur before quality of care is 
seriously affected. 

2.1.2 Studies and Relationships 

Four utilization studies are presented in this chapter: 

• PSRO Impact on Medicare Hospital Utilization 
Study (Medicare Impact Study) 

• PSRO Impact on Specific Diagnoses and Pro- 
cedure Category Study (Diagnosis- and Proce- 
dure-Specific Impact Study) 

• Medicaid Hospital Utilization Rate Study (Medi- 
caid Rate Study) 

• Case-Mix Adjusted Length of Stay Analysis (Case- 
Mix Adjustment Study) 

The Medicare impact study and the diagnosis- and 
procedure-specific impact study are evaluative studies. 
The Medicaid rate study and case-mix adjustment 
study are more developmental in nature. 

The Medicare impact study (Section 2.2) is designed 
to estimate the impact of the PSRO program on Medi- 
care inpatient utilization rates as of 1978. It examines 
utilization rates in all PSRO areas and is based on the 
most extensive source of Medicare data available (the 
100 Percent Medicare Claims File). As such, it is the 
most important utilization study in the evaluation and 
provides the basis for the benefit estimates used in 
the benefit-cost analysis presented in Section 3. 



' F. E. Browning and L. Coump, "Report to the Patient Care 
Planning Care Council on a Bed Utilization Study," 
Rochester, N.Y.: Council of Social Agencies, 1969. 

-' P. M. Gertman and P. M. Bucher, "Inappropriate Hospital- 
Bed Days, and Tlieir Relationship to Length of Stay 
Parameters" (presented at the Medical Care Section, 99th 
meeting of the American Public Health Association, 
Minnesota, October 1977). 

A. Quenldo, "Indirect Evaluation of a Randomized Sample 
of the General Pactitioner's Work (1954)," in The Efficiency 
of h/ledicaf Care, H. H. Leiden, N. V. Stenfen Knoese 1963. 

' J. D. Restuccia and D. C. Holloway, "Barriers to Appropriate 
Utilization of an Acute Facility," t\/ledical Care, July 1976. 

' A. Rosser, "The Reliability and Application of Clinical 
Judgment In Evaluating the Use of Hospital Beds," t^edical 
Care, January 1976. 

"J. G. Zimmer and E. S. Groomes, "An Observer Reliability 
Study of Physicians' and Nurses' Decisions In Utilization 
Review of Chronic Care Facilities," Medical Care, January- 
February 1969. 

• J. G. Zimmer, "Length of Stay and Hospital Bed 
Misutlllzation," hAedical Care, May 1974. 



16 



The 1978 PSRO evaluation determined that in 1977 
the PSRO program resulted in a relative reduction in 
Medicare beneficiary utilization of approximately 1.5 
percent. While this effect might be due to a general 
decrease in utilization, it might also be attributable to 
changes in hospital utilization for a specific subset of 
diseases. Using a sample of 20 percent of f\/ledicare 
hospital claims, the diagnosis- and procedure-specific 
impact study (Section 2.3) examines the impact of the 
PSRO program in 1977 on a number of procedures and 
diagnoses for which there is evidence of excessive 
lengths of stay or excessive admissions; these results 
are compared to aggregate findings and to results ob- 
tained from the analysis of a number of diagnoses and 
procedures for which there is no strong a priori ex- 
pectation of PSRO impact. 

The two studies mentioned above are concerned 
with PSRO impact only on Medicare utilization. PSRO 
impact on Medicaid utilization is, as yet, unknown due 
to the absence of a uniform national data base. The 
Medicaid rate study (Section 2.4) documents the steps 
necessary to develop consistent Medicaid hospital 
utilization rates and the potential for using Medicaid 
rates in evaluating PSRO performance. While the study 
is primarily developmental, limited descriptive results 
are presented based on four PSRO case studies which 
examine a combination of PSRO Medicaid discharge 
and State Medicaid agency data. A strong case cannot 
be made for PSRO impact on Medicaid beneficiary 
utilization, however, since no non-PSRO Medicaid dis- 
charge data are available for comparison (either inac- 
tive PSROs or pre-PSRO observations in active PSRO 
areas). 

The final study (Section 2.5) also is a developmental 
analysis, based on two case-mix adjusted indices of 
hospital utilization. The first index measures changes 
in the relative severity of illnesses of patients admitted 
by hospitals in active PSRO areas, either over time or 
across PSROs; the second measures changes in aver- 
age length of stay (ALOS) after adjusting for variations 
in case-mix severity. Both indices are based on a 
limited data base: namely, PSRO Hospital Discharge 
Data Set (PHDDS) data tapes received from the rela- 
tively small number of PSROs (12) for which tapes of 
acceptable quality are available for all of 1977 and 
1978. While the results cannot be said to be represen- 
tative of the PSRO program at large, they do indicate 
a marked trend toward increased case-mix severity and 
reduced length of stay among the twelve PSROs 
studied. Whether these findings are unique to those 
PSROs or reflective of a nationwide trend remains to 
be examined.* 

2.2 PSRO Impact on Medicare Hospital 
Utilization Study (Medicare Impact 
Study) 



'Also at issue is the possibility that these findings might be 
more indicative of changes in discharge reporting than of 
true changes in utilization. 



2.2.1 Objectives 

The PSRO program legislation was passed in 1972 
and the first conditional PSROs implemented review in 
1974. The number of active PSROs has since increased 
steadily and by June 1978 there were 108 PSROs ac- 
tively reviewing inpatient care for Federal beneficiaries. 
The main objective of this study is to determine what 
impact, if any, the PSRO program has had in reducing 
inpatient hospital utilization. As with the other utiliza- 
tion studies, the primary utilization variables are the 
days of care per 1000 Medicare enrollees (DOC), the 
number of discharges per 1000 Medicare enrollees 
(DISC), and the corresponding average length of stay 
(ALOS). 

The primary study hypotheses are as follows: 

1. There has been a reduction in DOC in active 
PSRO areas relative to inactive areas. 

2. There has been a reduction in DISC in active 
PSRO areas relative to inactive areas. 

3. There has been a reduction in ALOS in active 
PSRO areas relative to inactive areas. 

These hypotheses are tested both nationally and 
regionally. A series of secondary hypotheses is also 
tested. These hypotheses are developed below (Sec- 
tion 2.2.3.2) as an aspect of the development of the 
study's regression model. 

The remainder of this section is presented in six 
subsections. The first discusses data sources. This is 
followed by a description of study methods including: 

(1) the definition of active and inactive PSRO areas; 

(2) an outline of the descriptive statistics presentation; 

(3) development of the study regression model; and (4) 
the impact calculation methodology required to trans- 
form the regression results into estimates of the num- 
ber of days of care saved due to PSRO review. Study 
findings are then presented in three sections: (1) 
descriptive statistics, (2) forced order regression 
analysis, and (3) impact calculations. The section 
closes with a discussion of these results. 

2.2.2. Data Sources 

1 . 100 Percent Medicare Claims File: Every hospital 
admission for which a Medicare claim is made is 
entered into this file. It is a complete count of dis- 
charges and days of care. The address of the provider 
is also included so that discharges and days of care 
can be aggregated to the PSRO level. Data were used 
for the years 1974, 1975, 1976, 1977, and 1978 for 
which there were 7.1 million, 7.3 million, 7.7 million, 
8.1 million, and 8.5 million acute care discharges 

of aged Medicare beneficiaries respectively. 

2. Ttie Medicare Master Enrollment File: This file 
contains records for all Medicare beneficiaries in the 
United States. The use of this file and the 100 Percent 
Claims File enables the calculation of rates such as 
days of care per 1000 beneficiaries. The number of 
aged Medicare enrollees in 1974, 1975, 1976, 1977, and 
1978 were 22.0 million, 22.5 million, 22.9 million, 23.5 
million, and 24.0 million respectively. 



17 



3. The Medicare Provider of Service File: This file 
contains information on inpatient providers; i.e., hos- 
pitals qualified for Medicare reimbursement. By ex- 
amining the date when each provider came under 
binding PSRO review it is possible to estimate the level 
of review (percent of all discharges under review) for 
any PSRO area for any time period in the study and 
the length of time that review has been performed. 

4. Ttie Master Facility Inventory: This data base, 
provided by the National Center for Health Statistics, 
contains information on supply factors, such as the 
number of short stay hospital beds, which may affect 
Medicare utilization rates independently of any PSRO 
effect. 

5. The Area Resource File: This data base, main- 
tained by the Bureau of Health Manpower, contains 
economic, demographic, and health resource data by 
county. The file includes such items as population 
density, number of families with yearly incomes below 
$5,000, and number of physicians per 1,000 population. 

2.2.3. Methods 

2.2.3.1 DEFINITION OF ACTIVE AND INACTIVE 
AREAS 

The categorizing of PSRO areas into active and in- 
active areas is critical to the evaluation, since the 
evaluation is essentially a comparative analysis com- 
paring areas with active PSROs to those without. Dur- 
ing the 1977 and 1978 PSRO evaluations, considerable 
effort was expended in developing an appropriate 
definition for active PSRO areas, i.e., those areas con- 
ducting some level of PSRO binding review. 

Sensitivity analyses suggested that PSROs which 
initiate binding review of at least one hospital early 
in the transition period have an impact upon hospital 
use rates very much like PSROs with a high proportion 
of hospitals under review for the entire year. This find- 
ing appears to result from the fact that once PSROs 
initiate binding review in one hospital, the remaining 
hospitals rapidly follow suit. 

In the 1979 PSRO evaluation, PSROs are considered 
active if they had been performing review for at least 
half of the study year (i.e., prior to July 1978). Based 
on this definition, there are 108 active PSROs and 81 
inactive PSROs in the data base. Since all data are 
aggregated to the PSRO level, the data base includes 
a total of 189 observations.^ 

2.2.3.2 DESCRIPTIVE STATISTICS 

Descriptive analyses are presented for a variety of 
outcome and independent ("adjustment") variables. 



Outcome variables for active and inactive PSROs are 
compared at the national and regional levels (see Sec- 
tion 2.2.4) to show differences in utilization over time. 
Limited inferences concerning PSRO impact are drawn 
from these comparisons. Measures of independent 
variables are presented in order that differences be- 
tween active and inactive PSROs may be observed. 
These observed differences are related to the pos- 
sibility that active PSROs might represent a self- 
selected group with different features from inactive 
PSRO areas ^°; this possibility appeared to be borne 
out only in the South. 

2.2.3.3 DEVELOPMENT OF THE REGRESSION 
MODEL 

The general model underlying the analysis is of the 
form: 

n m 

Y = a + 2 b A + cP + dR + 2 ejij 
1=1 1=1 

where: 

Y = a vector of observations on the study's outcome 

(dependent) variables 
a = the constant term 

bi . . . b,„ c, d and e, . . . e^ = unstandardized 

regression co- 
efficients corre- 
sponding to a 
class of inde- 
pendent vari- 
ables 
Ai ^ a series of adjustment variables (variables, 
other than PSRO impact, which might affect 
hospital utilization) 
P =: the PSRO program variable 
R ^ a special program variable representing area 

rate-setting activities 
ij- ^ a series of interaction terms 

The development (specification) of this general model 
involves five steps: (1) selection of outcome variables; 
(2) selection of adjustment variables; (3) design of 
program variables; (4) inclusion of the interaction be- 
tween the PSRO program variable and other independ- 
ent variables; and (5) selection of an estimation tech- 
nique. These procedures are discussed below. 

Selection of Outcome Variables: As discussed in 
Sections 2.1.1 and 2.2.1, DOC, DISC, and ALOS are 
the outcome (dependent) variables selected for study. 
These variables may be explored either nationally or 
regionally; that is, the impact of the PSRO program 
may be estimated either at the national level or at the 
regional level. As noted earlier, the data are aggre- 



' Three PSROs have been omitted from the analysis due to 
data consistency problems: Kansas, Puerto Rico, and the 
Virgin Islands. In addition, eight Los Angeles PSROs were 
aggregated Into one PSRO and five Massachusetts PSROs 
were aggegated to one PSRO in order to compute 
utilization rates. 



^"Self-selection would bias the results if there were char- 
acteristics which differed between the active PSROs (a 
self-selected group) and the inactives, that would pre- 
dispose the active group toward greater reductions in days 
of care. This would be the case, for example, if medical 
societies in active PSROs were more committed to utiliza- 
tion review than those in inactive PSROs. 



18 



gated at the PSRO level with each PSRO representing 
an observation. The analysis is based on 189 PSRO 
areas (observations). 

To construct the basic outcome variable DOC for a 
given PSRO, two measurement issues had to be 
addressed. The first issue pertained to the effect of 
patient migration across PSROs, the second to a short- 
fail in the number of 1978 claims processed. 

Patient Migration. The utilization data employed in 
this study are provider-based, that is, they represent 
care provided in the hospitals in each PSRO area. 
However, not all people living in a PSRO area receive 
care in that area. Patient migration could cause one to 
overestimate the denominator of the days of care rate 
in a given PSRO area (if patients received care else- 
where) or to underestimate it if hospitals in a given 
PSRO area treated significant numbers of non- 
residents. 

Using an adaptation of a method proposed by 
Bailey," the population at risk for a given PSRO was 
estimated by allocating portions of Medicare enroll- 
ment from all PSRO areas based on each individual 
PSRO's contribution (as a percent of its total dis- 
charges) to the patient load in the given PSRO area. 
A detailed description of this procedure is provided 
by Deacon and Lubitz.^^ 

Shortfall in Processed Claims. A shortfall in claims 
records can affect estimated utilization rates. The 100 
percent file used in this study is updated as of the 
end of the first calendar quarter of 1979. Not all 1978 
claims were processed by that date, however, since 
inpatient stays which are unusually long, or unusually 
complex, take somewhat longer to process and may 
not be entered into the claims file until the following 
calendar year. Indeed, delayed discharge reports have 
been shown to have a longer average length of stay.^^ 
Thus, as later updates are made to the file in follow- 
ing years, the calculated average length of stay tends 
to increase. 

The days of care for 1978 were estimated by com- 
puting the ALOS of those discharges that were re- 
ported, and multiplying the computed ALOS by the 
number of 1978 admissions. (The figure on admissions 
is taken from the Medicare query file i* and presents no 
shortfall problems.) This procedure produces an esti- 
mate of DOC which is somewhat low because of the 



" N. T. J. Bailey, "Statistics in Hospital Planning and Design," 
Applied Statistics, November 1965, pp. 146-157. 

"R. Deacon and J. Lubitz, "Development of Hospital 
Utilization Measurements for PSRO Areas." (Presented at 
the Public Health Conference on Records and Statistics, 
Washington, D.C., June 5, 1978). 

'='G. S. Chulls, "Utilization of Short-Stay Hospitals Under 
Medicare 1968-71." Health Insurance Statistics, U. S. 
DHEW, SSA, ORS, DHEW, Pub. No. (SSA) 75-11702. 

"Whenever a Medicare patient Is admitted to a hospital, the 
hospital queries the Medicare program to determine 
eligibility, coinsurance, etc. The query file thus Is a count 
of all admissions. It does not, however, contain date of 
discharge and therefore cannot be used, by itself, to 
calculate days of care. 



underestimate of ALOS. However, the consequent 
downward bias is not an analytical problem if it is 
constant across all PSRO areas or if the extent of the 
bias is randomly distributed. Analyses indicate that bias 
does not appear to be a problem. ^^ 

Selection of Adjustment Variables and Development 
of Related Hypotheses: The model theoretically should 
include all variables which might conceivably affect 
Medicare beneficiary utilization. The inclusion of such 
explanatory factors is necessary in order to avoid con- 
founding the PSRO impact on utilization with the pos- 
sible effect other explanatory factors might have. If 
the model is "misspecified," the PSRO variable may 
reflect the causal influences of omitted explanatory 
factors. The set of factors included in the model to 
account for causal influences other than that of the 
PSRO program are called adjustment variables. 

The most important adjustment variable is the base 
value of the dependent variable. The year 1974 was 
chosen as the adjustment base year because this is 
the earliest pre-PSRO program year for which data 
exist; that is to say, the later years are contaminated 
with "transition" effects associated with the develop- 
ment of the PSRO program. The inclusion of baseline 
variables accounts for differences in levels of utiliza- 
tion across PSROs before PSRO review was initiated. ^^^ 

The use of a base measure has one other important 
implication. It enables the model to represent rate of 
change in utilization rates between 1974 and 1978. A 
true rate-of-change model was not used (i.e., the de- 
pendent variable was not specified as Yig.g — Yig-4) 
because the benefit measure ultimately used in the 
benefit-cost analysis (Section 3) is PSRO impact as 
of 1978, not the PSRO impact over the years 1974 to 
1978. Also, model specfication in this form eliminates 
potential covariance bias between the base rate and 
the change." 

The next group of adjustment variables was selected 
in order to isolate PSRO impact from other influences 



"'To investigate the possibility that bias might present a 
problem, the 100 Percent Medicare Claims File Study 
conducted In 1978 using 1977 data was replicated using 
1977 data updated as of the first quarter of 1979. Most of 
the previous year's shortfall was captured in this new 
update. Last year's results were confirmed indicating that 
the shortfall had no overall effect on the 1978 analysis. A 
second analysis compared 1977 total days of care from 
the 1978 and 1979 updates. The difference between the 
two updates (i.e., the shortfall) was analyzed by PSRO 
activity, region of the country and other possible 
explanatory variables; no systematic bias was found. In 
particular, shortfall differences were observed to be 
randomly distributed across active and Inactive PSRO areas 
indicating that shortfall does not bias impact estimates. 

'"Note that each dependent variable has its own unique 
base measure, i.e., the DOC base measure Is DOCwti, the 
DISC base measure Is DISCi3T). etc. 

'■ For a technical discussion of this issue see Lee J. 
Cronbach and Lita Furby, "How should we measure 
change — or should we?" Psychological Bulletin, Vol. 74, 
No. 1, pp. 68-80. 



19 



on utilization outcome variables. A brief discussion of 
some of those influences follows. 

The health care delivery system is recognized to be 
a complex system. The PSRO program is only one of 
the factors which influence health care utilization. 
Previous analyses have classified variables likely to 
have an effect on the use of health care as either 
supply variables or demand variables. The specification 
of the 1979 study's adjustment variables is influenced 
by the literature cited below, as modified by data 
availability. 

The supply of medical resources has been shown 
to influence hospital utilization. Research has shown 
that the supply of short-stay bedSj^^i^'^" the supply of 
physicians," and occupancy rates are all related to 
days of care rates. Increases in bed supply and 
higher occupancy rates tend to be associated with 
higher days of care rates. The effect of physician 
supply is somewhat mixed. Inpatient utilization tends 
to be positively correlated with the supply of hospital- 
based physicians but negatively correlated with the 
supply of general practitioners. 

The availability of some types of medical resources 
can act as substitutes for acute inpatient resources. 
In particular, long-term care beds represent an 
alternative mode of care. As the supply of long-term 
beds increases, the use of short-term beds is expected 
to decrease. 

A number of factors are known to affect the demand 
for hospital care. Health insurance, to the extent that 
it reduces the out-of-pocket cost of receiving health 
care, has been shown to increase the demand for 
hospitalization. 22. 23.24 All Medicare beneficiaries have 
the same Part A (hospital) coverage; some may have 
purchased private insurance to supplement their 
Medicare policies. However, in this study it was not 
possible to measure the extent of supplementary 
insurance coverage across PSRO areas. Since Medi- 
care covers most inpatient care and since this study 
is focused on Medicare utilization, not much is lost 
by omitting variations in insurance coverage from the 
analysis. 



" R. Deacon et al., "Analysis of Variations in Hospital Use 

by Medicare Patients in PSRO Areas, 1974-1977," Health 

Care Financial Review, Summer, 1979. 
"M. Feldstein, "Hospital Cost Inflation: A Study of Nonprofit 

Price Dynamics," The American Economic Review 65(5), 

December 1971: 853-872. 
^M. I. Roemer, "Bed Supply and Hospital Utilization: A 

Natural Experiment," Hospitals 35, (November 1, 1961): 

36^2. 
^ Deacon et al., op. cit. 
^ R. Anderson and J. Newman, "Societal and Individual 

Determinants of Medical Care Utilization," l\/lilbank 

Memorial Fund Quarterly 51 (Winter, 1973): 95-124. 
^ C. E. Ptielps and J. P. Newhouse, "Effect of Coinsurance: 

A Multi-Variate Analysis," Social Security Bulletin 35 

(June): 20-28, 1972. 
"J. P. Newhouse and C. E. Phelps, "Price and Income 

Elasticities for Medical Care Services," in M. Perlman, (Ed.), 

The Economics of Health and Medical Care. (London: 

Macmillan, International Economics Association, 1974). 



Other demand factors are related to demographic 
variations in the population. The evidence indicates 
that variations in age.^s income,^" and urban/rural 
location 27 can affect the days of care rates. The effect 
of age is well-known and documented: inpatient 
utilization is lowest for the youngest portion of the 
population and increases with age. In part, the Medi- 
care Impact Study controls for age differences by 
examining only aged (65 and over) Medicare bene- 
ficiaries. In addition, however the relative number of 
aged in a community could impact on utilization by 
varying the Medicaredemand as a percent of total 
hospital demand. Income also affects demand but not 
consistently. Low income persons generally have a 
lower demand for care; however when third party 
payors such as Medicare or Medicaid remove financial 
barriers to access to care the greater need for care 
among the poor causes their utilization to exceed 
that of those with higher incomes. Finally utilization 
rates are higher in urban areas than in rural areas. 
This may be due to a more limited access to care in 
rural areas or to a greater need for care in urban 
areas. 

Other factors not easily classified as demand or 
supply variables, can affect the utilization of hospitals. 
One of these factors is the presence or absence of 
state hospital rate review commissions. These com- 
missions, currently operating in ten states, function 
to control hospital budgets through regulation and 
control of reimbursement rates. However, it is possible 
that hospitals respond to such budgetary constraints 
by admitting more patients, extending lengths of stay 
or increasing service intensity. Any of these changes 
in utilization would enable hospitals to maintain 
revenues in spite of limits on per diem reimbursement. 
Limited research suggests this may in fact be 
happening. 28 

The previous discussion is not meant to be an 
exhaustive review of the health services research on 
factors influencing hospital utilization. It is a brief 
listing of some of the factors which should be 
accounted for when developing a series of hypotheses 
and a model of hospital utilization. 

In addition, there are numerous factors which vary 
by geographic region but which could not be 
measured and included in this analysis. Among these 
are climatological variations, differences in physician 
and hospital treatment modalities, and additional socio- 
economic considerations. To account for these varia- 
tions, a series of regional variables were entered into 



== B. J. Haught, "Utilization of Short-Stay Hospitals, 1977," 

Vital and Health Statistics," Series 13, (41) DHEW 

Publications (PHS) 79-155. 
=° Andersen and Newman, op. cit. 
" National Center for Health Statistics, "Current Estimates 

from the Health Inten/iew Survey: U. S. 1975," Vital and 

Health Statistics, Series 10, (115) DHEW Publication No. 

(HRA) 77-1543, March 1977. 
°'Abt Associates, "National Hospital Rate Setting Study: 

Research Objectives and Methodology," DHEW Contract 

No. 500-78-0036, 1979. 



20 



the model. Census regions are used; the states 
included in each of the four Census regions are listed 
in the following table: 



used in the presentation of findings (Section 2.2.5). 
The data source refers to the data files listed in 
Section 2.2.2. 



A LISTING OF STATES BY CENSUS REGION 



LISTING OF ADJUSTMENT VARIABLES 



Region 



States Included 



Code 



Variable 



Source 



Northeast 



North Central 



South 



West 



Maine, New Hampshire, Ver- 
mont, Massachusetts, Rhode 
Island, Connecticut, New York, 
New Jersey, Pennsylvania 

Michigan, Ohio, Indiana, Illinois 
Wisconsin, Minnesota, Iowa, 
Missouri, North Dakota, South 
Dakota, Kansas, Nebraska 

Delaware, Maryland, District of 
Columbia, Virginia, West Vir- 
ginia, North Carolina, South 
Carolina, Georgia, Florida, 
Kentucky, Texas, Tennessee, 
Alabama, Mississippi, 
Arkansas, Louisiana, Oklahoma 

Montana, Idaho, Wyoming, 
Colorado, New Mexico, 
Arizona, Utah, Nevada, 
Washington, Alaska, Oregon, 
California, Hawaii 



It should be kept in mind that geographic region 
per se should not theoretically have a major impact 
on health care utilization, except for some climatic 
effects on disease and disability incidence and the 
indirect effects of regional geographic characteristics 
on economic development which in turn affects the 
availability of health care services. However, experi- 
ence has shown that there are in fact regional varia- 
tions in health care utilization,^" supply of health 
facilities and manpower ^o and that regional effects on 
utilization do exist apart from demographic, economic, 
and health care supply characteristics. ^i Therefore, 
given the historical fact that regional differences do 
exist, there is sufficient justification to include region 
in the analysis. 

National data which can be aggregated at the PSRO 
level are not available for all possible supply and 
demand factors which might affect Medicare bene- 
ficiary hospital utilization. The adjustment variables 
for which data were available are listed below. The 
adjustment variable code represents the variable name 



^NCHS, Vital and Health Statistics, Series 10 and Series 13. 
'" USDHEW, Healtti; United States: 1978 DHEW Publication 

No. (PHS) 78-1232. 
='• USDHEW, HCFA, PSRO 1978 Program Evaluation, DHEW 
Publication No. HCFA-03000 January, 1978. 

• Deacon et al., op. cit. 

• Rothberg et al., Medicare Utilization in ttie U.S.: PSRO 
and Regional Impacts, University Health Policy 
Consortium 



BASE 


1974DOC, DISC, orALOSfor 
aged Medicare beneficiaries. 


100%, MEF 


AGED 


Number of persons aged 65 
and over/1000 population 
(change, 1976-1974). 


ARF, MEF 


BEDS 


Number of short-stay beds/ 
1000 population (change, 
1976-1974). 


ARF 1 


DEN 


Population density per 
square mile (1976). 


ARF 


DOCLD 1 


Number of Medicare days of 
care/1000 days of care for 
all persons. 


100% MFI 


MD 


Number of physicians/1000 
population (change, 1976- 
1974). 


ARF 


LTC 


Number of Medicare-certified 
long-term care beds/ 1000 
Medicare beneficiaries 
(1978). 


PCS, MEF 

1 


OCC 32 


Occupancy rate (1976). 


MFI 


POV 


Number of families with less 
than $5000 income per 1000 
population. 


ARF 

1 


TEACH 


Number of beds in teaching 
facilities/100 total short- 
stay beds. 


POS 


REG 


Census region variables. 


BOC 



As in the 1978 evaluation, several of the adjust- 
ment variables have been entered in change form; 
i.e., as the difference between their value in the most 
recent year (1976) and their value in the base year 
(1974). The reason for this is that the model is essen- 
tially a rate of change model: although the dependent 
variable is expressed in 1978 terms (to facilitate 
benefit-cost analysis), its base value appears on the 
right-hand side of the equation. It would be inappro- 



-The occupancy and percentage of Medicare days of care 
variables' effect on Medicare days of care may appear to 
be tautological. That is to say, by definition as occupancy 
and Medicare days of care load factor increases, so 
should total Medicare days of care. However, these 
variables are entered as lag variables. Therefore, their 
impact is a delayed one and not indicative of an alternative 
measure of the same phenomenon in the same year. 

■ 100% = 100 Percent Medicare Claims File 
ARF =r Area Resource File, Bureau of Health Manpower 
MFI = Master Facility Inventory, National Center for 

Health Statistics 
POS = Medicare Provider of Service File 
MEF = Master Enrollment File — Medicare 
BOC = Bureau of Census regional groupings 



21 



priate, however, to express all of the adjustment 
variables in change form since changes in those 
variables might correlate with the impact of PSRO 
review. If this were so, the includsion of those 
variables in change form could tend to reduce the 
apparent effect of PSRO review. This argument applies 
most particularly to percent of total days of care due 
to Medicare and to the occupancy measure. For 
example, if PSRO review were to reduce Medicare 
DOC within a given PSRO, one would expect a simul- 
taneous decline in the occupancy rate and a smaller 
proportion of days of care due to Medicare (assuming 
all other factors remained constant). For this reason, 
these adjustment variables (occupancy rate and per- 
centage of total days of care due to Medicare), plus 
several others, were entered in terms of their most 
recent values. On the other hand, the adjustment 
variables related to physicians, short-stay hospital 
beds, and proportion of aged population were entered 
in change form, i.e., the numeric difference between 
the most recent value of that variable and its value in 
1974. 

While the primary intent of this study is to explore 
the relationship of PSRO review to utilization, 
hypotheses concerning individual adjustment variables 
can also be generated. Specifically, it is expected 
that hospital utilization of aged Medicare beneficiaries 
will tend to decrease: 

• With decreases in the total occupancy rate (OCC); 

• With decreases in the proportion of the population 
below the poverty level (POV); 

• With decreases in the population density (DEN); 

• With decreases in the change in short-stay bed 
supply (BEDS); 

• With increases in the long-term care bed supply 
(LTC); 

• With decreases in the change in physician supply 
(MD); 

• With decreases in the change in elderly population 
(AGED); 

• With decreases in the percent of all days of care 
that are due to Medicare (DOCLD); 

• With decreases in the percent of short-stay beds 
which are in teaching facilities (TEACH); and 

• Will vary across census regions (REG). 

Design of Program Variables: The PSRO program 
variable is the focus of the study. The hospital rate 
review commission variable has been added to test 
whether rate review had an impact on the outcome 
variables. 

In the 1978 evaluation, numerous forms of the PSRO 
variable were tested. The form eventually selected, and 
used this year as well, was called PSRO longevity. 
The longevity variable included in this year's model 
is defined as the number of months of PSRO activity 
prior to July 1, 1978. If the regression coefficient for 
PSRO longevity (Section 2.2.3.3) is negative and 
significant, the null hypothesis of no program impact 
will be rejected. A favorable PSRO impact will be 
inferred. 



The hospital rate review commission variable is a 
dummy variable equal to "1" when rate review is 
present in a PSRO area and "0" when no rate review 
is present. The state rate review commissions included 
in the analysis are New York, Connecticut, New Jersey, 
Rhode Island, Wisconsin, Washington, Indiana, 
Colorado, Massachusetts, and Maryland. These were 
selected from listings found in abstracts of state- 
legislated hospital cost containment programs 
published by HCFA in 1978. 

Inclusion of Interaction Terms: The remaining 
variables entered into the model are the interaction 
terms. These were entered to test for the possibility 
that PSRO impact varies according to some other 
characteristic. For instance, the "PSRO by base rate" 
interaction term tests for the possibility that the PSRO 
impact might be different in areas with higher or lower 
days of care rates in 1974. Similarly, the "PSRO by 
region" interaction term tests for the possibility that 
PSRO impact may vary across regions.^* 

It should be emphasized that interaction terms are 
not "catch all" variables included only to account for 
residual variance in the model. If it were the case that 
all explanatory variables were perfectly independent 
of each other (i.e., zero correlation among these 
variables), there would be no necessity for including 
interaction terms. This is rarely the case, however, in 
non-experimental research. For the PSRO evaluation 
in particular, interaction terms can be very useful. If 
it happens that PSRO effectiveness varies by some 
demographic or health care supply characteristic, then 
there are implications for the program itself. For 
example, PSRO effectiveness could interact with 
Medicare-certified long term care beds. Such an 
interaction might suggest that PSRO effectiveness is 
constrained by the supply of long term care beds and 
that PSRO effectiveness could be improved by 
increasing (or decreasing) the supply of long term 
care beds.^"^ 

Selection of Estimation Technique: Multiple regres- 
sion is the estimation technique used in this study. 
This technique allows one to test for the effect of the 
PSRO program on an outcome variable while con- 
trolling for, or holding constant, the effects of other 
variables. One variant of the regression model, forced 
order multiple regression, is used in this study as it 
was in the 1978 and 1977 evaluations. 

Forced order multiple regression uses the variables 
developed above in the following fashion. For a given 
outcome variable, the independent variable sets (base 
rate, adjustment, program, and interaction) are entered 
in a predetermined order so as to build, in a stepwise 
fashion, a multiple regression equation for explaining 
variations in the outcome variable. Variables are 



•"The region variables themselves test for variation in the 

dependent variables across regions. 
■"^Thls Is a hypothetical example. In the analysis Itself, the 

Interaction of PSRO reviews and Medicare-certified long 

term care beds was non-significant. 



22 



TABLE 6 

Total Days of Care per 1000, Total Discharges per 1000 and 

Average Length of Stay for Aged Medicare Beneficiaries by 

PSRO Area, 1974 to 1978: U.S. Totals " 



1974 



1975 



1976 



1977 



1978 



Percent Change 
1974-1978 



Days of Care/1000 
All Areas (N=rl89) 
Active (N=108) 
Inactive (N=81) 

Discharges/1000 
All Areas (N=189) 
Active (N=108) 
Inactive (N=81) 

Average Length of Stay 
All Areas (N=189) 
Active (N=108) 
Inactive (N=81) 



3641 
3652 
3625 


3604 
3613 
3592 


3698 
3712 
3679 


3647 
3644 
3651 


3667 
3671 
3662 


+0.7% 
-hO.5% 
+1.0% 


318 
313 
326 


323 
317 
332 


334 
328 
341 


336 
329 
344 


344 
337 
352 


+8.0% 
+7.8% 
+8.1% 


11.6 
11.8 
11.2 


11.3 
11.6 
10.9 


11.2 
11.5 
10.9 


11.0 
11.2 
10.7 


10.8 
11.0 
10.5 


—6.9% 
—7.3% 
—6.7% 



entered in linear form.^*' Tests of significance are made 
at each step of the analysis so that the significance of 
the incremental contribution of each set of variables 
can be assessed. A conservative test of significance 
is made for the PSRO variable in that the PSRO 
variable is entered into the model only after the 
regression equation has controlled for all adjustment 
variables. 

2.2.3.4 IMPACT CALCULATIONS 

The primary purpose of the regression analysis is 
to determine if PSROs impact upon Medicare bene- 
ficiary utilization. Significant PSRO longevity coeffi- 
cients (with negative signs, indicating reductions in 
utilization) suggest favorable impact. Favorable impact 
in and of itself, however, is not sufficient information 
to support a benefit-cost analysis of PSRO concurrent 
review. Impact information must be transformed into 
an estimate of the actual number of days "saved" by 
PSRO review activity. This estimate can then be used 
to develop the dollar measure of benefits required by 
the PSRO concurrent review benefit-cost analysis 
(Section 3). Section 2.2.6 builds upon the regression 
findings to present estimates of the number of days 
of care reduced due to PSRO concurrent review 
activity. 

2.2.4 Findings: Descriptive Analyses 

2.2.4.1 NATIONWIDE TOTAL OUTCOME VARIABLES 

Table 6 presents annual DOC, DISC, and ALOS data 
for aged Medicare beneficiaries in active and inactive 



PSRO areas for the years 1974 through 1978. As can 
be seen, days of care rates did not change greatly 
over the five-year period. For active PSROs, DOC 
increased from 3,652 to 3,671 , an increase of about 
0.5 percent. In inactive areas, the increase was from 
3,625 to 3,662, an increase of approximately 1 percent. 
While the unadjusted data thus indicate that DOC 
increased at a slower rate in active PSRO areas than 
in inactive areas the difference is slight (0.5%); 
moreover, the comparison does not control for other 
factors which might affect the days of care rates. 

Examination of the discharge rate and ALOS series 
also indicates small differences between active and 
inactive PSRO areas. Although inactive areas had a 
slightly higher discharge rate in 1974 than did active 
areas (326/1000 and 313/1000, respectively), the 
increase was slightly greater for inactives (8.1 %) than 
for actives (7.8%) in 1978. Actives, conversely, had 
a longer ALOS in 1 974 than inactives (1 1 .8 and 1 1 .2 
days, respectively); again, however, there was not 
much difference between active and inactive areas in 
terms of percentage change over the next four years. 

A simple examination of nationwide use rates 
between active and inactive PSRO areas does not lead 
to strong conclusions about the effectiveness of 
PSROs. Differences in use trends do exist and are in 
a favorable direction, but they are small. However, the 
examination of changes in utilization rates fails to 
account for the adjustment variables mentioned 
earlier which could impact upon utilization rates and 
thereby mask a true PSRO effect. 



'"The analysis is performed with variables measured in their 
observed form. The regression was re-estimated with the 
variables transformed into logarithmic units. The results 
are similar. 



"■' Figures in this and following tables may not coincide with 
figures in the 1978 evaluation. This is due to the exclusion, 
from this year's data base, of inappropriate hospitals which 
were included in last year's study. This issue is discussed 
further in Section 2.2.7. 



23 



2.2.4.2 CENSUS REGION OUTCOME VARIABLES 

This subsection presents data on outcome variables 
for the four census regions: Northeast, North Central, 
South, and West 

Northeast Region: Table 7 presents annual DOC, 
DISC, and ALOS data series for aged Medicare bene- 
ficiaries by active and inactive PSRO areas in the 
Northeast region. Day of care per 1000 beneficiaries 
rose by 5.6 percent over the five-year observation 
period. However, the increase was much less in active 
PSRO areas than in inactive areas. In active areas, 
the DOC rose from 3,888 in 1974 to 4,030 in 1978, an 
increase of 3.7 percent. In inactive PSRO areas, the 
increase was from 3,781 to 4,244, an increase of 12.2 
percent. 

The increase of DOC was due to a large percentage 
increase in discharge rate, greater than the percentage 
decrease in ALOS. Overall, the discharge rate 
increased by 11 percent in the Northeast although the 
increase was slightly less for active than for inactive 
PSRO areas (10% and 13%, respectively). The main 
difference between active and inactive PSRO areas 
was in ALOS. in 1974, active PSRO areas had an 
ALOS slightly higher than inactive PSRO areas (13.7 
days and 13.6 days, respectively). However while the 
ALOS in active areas decreased by 6.6 percent over 
the next four years, the ALOS in inactive areas 
remained unchanged. 

As in the case of the nationwide figures presented 
earlier, active PSRO areas in the Northeast region 
showed a lower percentage increase in utilization than 
did the inactive areas. Again, however, the differences 
are not adjusted for other possible sources of 
explanation. 

North Central: Utilization rates for aged Medicare 
beneficiaries in the North Central region are presented 



in Table 8. Unlike the Northeast, the days of care per 
1000 decreased during the five-year observation period. 
For active PSRO areas, the decrease was 3.0 percent 
(4,147 to 4,024) whereas for inactive PSRO areas there 
was a smaller decrease of 1.2 percent (3,883 to 3,837). 
The larger decrease in DOC active areas was due to a 
smaller percentage increase in DISC (6.9% vs. 8.5%) 
and a larger percentage decrease in ALOS (9.9% vs. 
9.3%) in the active than in the inactive areas. 

South: Medicare utilization rates for the South region 
are presented in Table 9. The South shows trends in 
utilization between active and inactive areas, which 
differ from those in the Northeast and North Central 
regions. DOC has risen by 4.7 percent in active PSRO 
areas while declining slightly (0.4%) in inactive areas. 
DOC for both areas combined has risen by 1.6 percent. 

In 1974, the discharge rate in active PSRO areas was 
10 percent lower than in inactive areas (310 and 344, 
respectively). Since then, the discharge rate has risen 
rnore rapidly in active areas (12.3%) than in inactive 
areas (7.6%). Both active and inactive have shown 
decreases in ALOS but the decline has been greater 
in the inactive areas (6.7% or active vs. 7.5% for 
inactives). 

West: Finally, the West region has had an overall 
decrease in DOC of 4.2 percent. As shown in Table 10, 
in active PSRO areas DOC decreased from 2,893 to 
2,748 (5%) while in inactive areas DOC decreased 
slightly, from 2,711 to 2,670 (1.5%). Active and inactive 
PSRO areas had identical ALOS in 1974 (9.0 days) and 
since that time experienced identical decreases (5.6%). 
The discharge rate, however, has increased less rapidly 
in active than in inactive PSRO areas (1.6% vs. 4.3%), 
thereby accounting for the greater decline in DOC in 
the active areas. 



TABLE 7 

Total Days of Care per 1000, Total Discharges per 1000 and 
Average Length of Stay for Aged Medicare Beneficiaries by 
Active and Inactive PSRO Areas, 1974 to 1978: Northeast 





1974 


1975 


1976 


1977 


1978 


Percent Change 
1974-1978 


Days of Care/1000 














All Areas (N— 46) 


3863 


3884 


4035 


4001 


4081 


+5.6% 


Active (N=35) 


3888 


3893 


4027 


3963 


4030 


+3.7% 


Inactive (N=11) 


3781 


3855 


4059 


4119 


4244 


+12.2% 


Discharges/1000 














All Areas (N— 46) 


284 


290 


302 


305 


315 


+11.0% 


Active (N=35) 


286 


291 


303 


306 


316 


+10.0% 


Inactive (N=11) 


278 


286 


299 


303 


313 


+13.0% 


Average Length of Stay 














All Areas (N— 46) 


13.7 


13.5 


13.4 


13.2 


13.0 


—5.1 % 


Active (N=35) 


13.7 


13.5 


13.4 


13.0 


12.8 


—6.6% 


Inactive (N=11) 


13.6 


13.5 


13.6 


13.6 


13.6 


0% 



24 



TABLE 8 

Total Days of Care per 1000, Total Discharges per 1000 and 

Average Length of Stay for Aged Medicare Beneficiaries by 

Active and Inactive PSRO Areas, 1974 to 1978: North Central 





1974 


1975 


1976 


1977 


1978 


Percent Change 
1974-1978 


Days of Care/1000 














All Areas (N=51) 


4007 


3953 


4019 


3937 


3925 


—2.0% 


Active (N=24) 


4147 


4091 


4157 


4045 


4024 


—3.0% 


Inactive (N^27) 


3883 


3830 


3896 


3841 


3837 


—1.2% 


Discharges/1000 














All Areas (N=51) 


338 


343 


354 


356 


364 


+7.7% 


Active (N=24) 


348 


352 


363 


364 


372 


+6.9% 


Inactive (N — 27) 


330 


336 


345 


348 


358 


+8.5% 


Average Length of Stay 














All Areas (N=51) 


11.9 


11.6 


11.4 


11.1 


10.8 


—9.2% 


Active (N=24) 


12.1 


11.7 


11.6 


11.2 


10.9 


—9.9% 


Inactive (N— 27) 


11.8 


11.4 


11.3 


11.1 


10.7 


—9.3% 



TABLE 9 

Total Days of Care per 1000, Total Discharges per 1000 and 

Average Length of Stay for Aged Medicare Beneficiaries by 

Active and Inactive PSRO Areas, 1974 to 1978: South 





1974 


1975 


1976 


1977 


1978 


Percent Change 
1974-1978 


Days of Care/ 1000 
All Areas (N=57) 
Active (N^23) 
Inactive (N=34) 


3621 
3633 
3613 


3594 
3630 
3571 


3687 
3751 
3643 


3670 
3756 
3611 


3680 
3802 
3597 


+1.6% 
+4.7% . 
—0.4% 


Discharges/1000 
All Areas {N=57) 
Active (N— 23) 
Inactive (N— 34) 


330 
310 
344 


337 
318 
351 


348 
331 
360 


353 
337 
364 


361 
348 
370 


+9.4% 

+12.3% 

+7.6% 


Average Length of Stay 
All Areas (N=57) 
Active (N=23) 
Inactive (N=34) 


11.1 
11.9 
10.6 


10.8 
11.6 
10.3 


10.7 
11.5 
10.2 


10.5 
11.3 
10.0 


10.3 

11.1 

9.8 


—7.2% 
—6.7% 
—7.5% 



TABLE 10 

Total Days of Care per 1000, Total Discharges per 1000 and 

Average Length of Stay for Aged Medicare Beneficiaries by 

Active and Inactive PSRO Areas for Years 1974 to 1978: West 





1974 


1975 


1976 


1977 


1978 


Percent Change 
1974-1978 


Days of Care/1000 
All Areas (N=35) 
Active (N— 26) 
Inactive (N==9) 


2847 
2893 
2711 


2743 
2779 
2639 


2807 
2842 
2705 


2723 
2745 
2657 


2728 
2748 
2670 


—4.2% 
—5.0% 
—1.5% 


Discharges/1000 
All Areas (N— 35) 
Active (N=26) 
Inactive (N=9) 


314 
319 
301 


314 
318 
302 


321 
325 
310 


317 * 

320 

309 


321 
324 
314 


+2.2% 
+1.6% 
+4.3% 


Average Length of Stay 
All Areas {N=35) 
Active (N=:.26) 
Inactive (N=9) 


9.0 
9.0 
9.0 


8.7 
8.7 
8.7 


8.7 
8.8 
8.7 


8.6 
8.6 
8.6 


8.5 
8.5 
8.5 


-5.6% 
-5.6% 
—5.6% 



25 



Several important aspects of utilization are suggested 
by an examination of Tables 7 through 10. First it is 
evident that discharge rates are increasing in all four 
Census regions while lengths of stay are declining, for 
both active and inactive PSRO areas. However, the 
overall utilization trend, as measured by the variable 
DOC, is distinctly different within each region. In the 
Northeast, utilization is increasing (since DISC is rising 
more rapidly than ALOS is declining) while in the 
North Central and West, DOC are declining; in these 
three regions, however, the trend is more favorable in 
areas which have active PSROs than in those which do 
not. In the South, however, utilization has risen in ac- 
tive areas while declining slightly in those which are 
inactive. 

These regional differences suggest that PSRO review 
takes place within the context of a multitude of other 
factors which influence utilization. The different re- 
gional trends suggest that these contextual factors do 
vary by region and that those variations affect PSRO 
performance. Thus the unadjusted data suggest that 
PSROs may be having a favorable impact in the North- 
east, North Central and West regions, but an unfavor- 
able impact in the South. The simple differences do not 
account for other factors which could affect utilization, 
and the raw figures do not indicate whether the ob- 
served differences are statistically significant. 



2.2.4.3 NATIONWIDE TOTAL ADJUSTMENT 
VARIABLES 

Table 1 1 presents the mean values of the adjustment 
variables for active and inactive PSRO areas. A com- 
parison of these values provides some indication of the 
extent to which active areas are unlike inactive areas. 
Large discrepancies would suggest that active PSROs 
are possibly "self-selected" and perhaps behaviorally 
different from inactive PSRO areas. While the regres- 
sion model is used to reduce the influence of such 
self-selection it is likely that the model does not con- 
tain all relevant causal variables (i.e., is misspecified). 
Comparability in observed variable levels, on the other 
hand, suggests that the non-observed variables might 
also be similar. If true, the argument for self-selection 
is weakened. Unfortunately, there is no way to com- 
pletely settle the issue since no direct observation of 
discrepancies across active and inactive PSROs is 
possible for unmeasured variables. 

As shown in Table 11, active PSRO areas tend to be 
more urban (higher population density) than inactive 
PSRO areas (2415/square mile and 572/square mile, 
respectively). Also, the number of Medicare-certified 
long stay beds per 1000 Medicare beneficiaries is 
statistically significantly greater in active PSRO areas 
than in inactive areas (23.0 and 17.2, respectively) as 
is the percent of short stay beds which are in teaching 



TABLE 11 
Mean Values of Adjustment Variables for Active and Inactive PSRO Areas 



All 

PSRO Areas 

(N=189) 



Active 
(N=108) 



Inactive 
(N=81) 



Change in Short Stay Beds/1000 

Persons (1974-1976) 
Change in Physicians/1000 

Persons (1974-1976) 
Change in Number of Persons 

Age 65 and Over per 1000 

Population (1974-1976) 
Population Density Per Square 

Mile (1976) 
Number of Days of Care 

Due to Medicare per 1000 

Total Days of Care 
Medicare Certified Long Term 

Care Beds/1000 Beneficiaries 

(1978) 
Poverty Families/1000 Persons 
Occupancy Rate 
Number of Short Stay Beds in 

Teaching Facilities per 

100 Short Stay Beds 



.06 
.10 
3.3 

1625 
342 

20.5 



110 
.74 
31 



.03 
.11 
2.8 

2415* 
340 

23.0* 



110 
.74 
35* 



.10 
.10 
3.9 

572* 
345 

17.2* 



110 
.73 
26* 



* Denotes characteristics for which there was a statistically significant difference (p < .05) between active and inactive 
PSRO areas. 



26 



facilities (35% as against 26%). Thus, at the national 
level, there is some indication that the selection proc- 
ess has resulted in a set of active PSROs which are 
systematically different from inactive areas: active 
areas are more urban, have more Medicare certified 
long stay beds and a greater percentage of short stay 
beds in teaching facilities. 

Although the active and inactive areas differ on these 
individual characteristics, a multivariate comparison of 
active and inactive differences was performed on the 
set of adjustment variables. ^ This comparison showed 
that for the adjustment variables as a group, there is 
no significant difference between active and inactive 
areas. Accordingly, there is reason to believe that 
self-selection, while apparently present to some extent, 
is not a dominant factor in the analysis. 

2.2.4.4 CENSUS REGION ADJUSTMENT VARIABLES 

Table 12 displays the adjustment variable means for 
active and inactive areas by region. The table shows 
that there are major differences between the regions 
in the values of these variables. In addition, within 
each region there are some differences between active 
and inactive areas. 

The Hotelling P statistic was calculated for each 
region to compare the differences between active and 
inactive areas. In only one region, the South, were the 
set of adjustment variables statistically different be- 
tween active and inactive areas. This suggests that 
self-selection may be a more important factor in the 
South than in the other regions. 

2.2.5 Results: Forced Order Regression Analysis 

2.2.5.1 U.S. TOTAL RESULTS: DOC 1978 

Table 13 presents the forced order regression model 
results for the study's primary impact variable, DOC. 
The sequence of entry into the stepwise regression is 
that outlined in Section 2.2.3.2: (1) base value of the 
dependent variable; (2) other adjustment variables; (3) 
region; (4) PSRO review (i.e., longevity); (5) rate re- 
view; and (6) interaction terms. The degrees of freedom 
column in Table 13 indicates how many variables are 
entered at each step. The increase in R= associated 
with each step indicates the degree to which a variable 
(or group of variables) contributes to the explained 
variation in the dependent variable. Each step's F- 
value and corresponding significance level indicates 
the degree to which that step contributes significantly 
to the explained variation. A conventional .05 p-value 
(confidence level) is used as the threshold of signifi- 
cance. The following text describes the influence of 
each contribution to explained variance. 



'The multivariate comparison was done with the Hotelling 
V statistic. It is the multivariate extension of the ordinary 
Student t-test for the difference of two sample means. The 
value of P in this comparison was 19.97 with a p value of 
.146. 



The DOC 1974 base rate variable accounts for over 
81 percent of the total variance in 1978 days of care. 
After accounting for this variance, the impact of the 
remaining variables is their effect on the differences 
between the 1974 and 1978 days of care rates. 

The adjustment variables, as a group, account for 
an additional 9.4 percent of explained variance; this 
increase in R- is significant at the .001 confidence 
level. However, not every individual adjustment variable 
has a statistically significant impact. Four of the ad- 
justment variables account for most of the increase in 
R"; they are bed occupancy rate, the change in short 
stay beds per 1000 persons, the change in the percent- 
age of aged, and the percentage of days of care due to 
Medicare. Of these variables, all but the change in 
percentage of aged are positively associated with total 
days of care. That is, as occupancy, the short day bed 
supply (1976 minus 1974), and the Medicare portion of 
all days of care increase, so do total days of care 
for Medicare beneficiaries. Changes in the percentage 
of aged, on the other hand, are negatively associated 
with total days of care. While this finding might appear 
paradoxical, it does not indicate that areas with a 
greater percentage of aged persons are more likely to 
experience a lesser number of days of care for the 
aged. Rather, it indicates that areas with a greater in- 
crease in the percentage of aged persons are more 
likely to experience such a decline. In fact, the number 
of aged persons may be stable or decreasing in some 
areas while the non-aged population is decreasing 
more rapidly. In such an area, the percentage of aged 
would tend to increase while Medicare DOC would 
tend to decline. 

Three dummy regional variables were then entered 
into the equation to test for a regional effect. The F 
ratio (5.63) was significant, confirming the earlier con- 
jecture that there are significant regional differences 
in DOC. 

The PSRO program variable was entered into the 
equation at the next step. Although the effect was to 
increase R^ by only 0.1 percent, the sign of the coeffi- 
cient is negative which indicates that the PSRO impact 
is in the favorable direction (i.e. reducing DOC). This 
effect was not significant at the .05 confidence level, 
however. It appears, then, that there is no statistically 
significant nationwide PSRO effect. However, as further 
analysis will indicate, this does not imply that there is 
no PSRO effect at all. Rather, it indicates that if a 
PSRO impact does exist it is not consistent across all 
PSRO areas. 

The next variable to be included in the equation was 
a measure of hospital rate review commissions. This 
was a test of whether rate review had an impact on 
DOC. As can be seen in Table 13, the rate review effect 
on DOC was insignificant. However, the sign of the 
regression coefficient (positive) indicates that the 
tendency is for days of care to increase in PSRO areas 
which also have rate review activities. 

The final step in the analysis of the stepwide regres- 
sion model consists of the interaction variables. As 
Table 13 Indicates, four types of interactions with 



27 



CM 



O 

'So 
« 

JS 
M 

o 
tc 

(O 

OL 

.1 
'*^ 
u 
ra 



•a 
c 

o> 

< 



UJ V 

-I i 

2 « 

M 



ra 

> 



< 



o 
ra 

D. 

E 
o 
u 



> 5? 



0) (O 
> CM 

■^ II 



« lO 

— ra CO 

<z 

> ^ 

~ CO 

" II 

ra 4 



0) CO 
> CJ 

o II 
<Z 



W 1^ 

— ra lo 

< 2 II 

<z 



0) J-- 

> f^ 

" II 

ra 4 



■* 

'*-* II 
o II 
<Z 



</) 1- 
— ra m 

<2 II 
< Z 



ra 4 
c Z 



in 

> CO 

"5 II 
<Z 



(O CO 

— ra ^ 

< £ II 
<z 



c 


m 


0) 


(!) 


E 


o 


(0 


ra 


3 


k. 




ffl 


< 


> 



1^ 

o 



I 

CM 

7 

00 



O) 



00 



CO 

o 



O 



CO 

o 



c» 

o 



00 

o 



o> 
o 



CO 

o 



o 



o 



a> 



CO 



CM 
CM 



ir> 



CO 
CO 



■* 
■^ 



CM 



CO 


00 


■* 


o 


lO 


CO 



O) 

r- 



O) 



o 

CO 



lO 

o 

CO 



CO 



00 
00 



CM 

o 



CO U) 
(O CM 



CO ■* 

(O CM 



T- 


00 


1^ ^ 


O 

CO 


o 


(O CM 



O 


CO 


O) 


U5 




CO 



in 


00 


O) 




CM 


CO 



CO 


O) 


CO 


CO 


(O 


CO 



CO 


^- 


CO 


CO 


CO 


CO 



CJ) 


>* 


in 


in 


in 


CO 



CM 
O 



CO 
CM 



00 
CO 



to 



U) 
CO 



in 


CM 


in 


rr 


"* 


CO 



o 

00 



o 

in 



00 

d 



<3) 



(O 



CO 

1^ 



m 

CM 



CM 
CM 



CM 

o 



CO 



O) 

o 



t- CO 

r^ CM 



CO f^ 

r^ CO 



CO o> 
h- CM 



t^ CM 



h- CO 



Tf CM 

r>- CO 



(O 


(O 


00 


00 


CM 


CO 


CM 





CO 


o 


o ■* 


in 


O) 


00 CO 


CM 







00 


.,_ 


CO 


1^ 


<o 


CO 


in 





CO 
00 



CO 



(O 

OJ 
CM 



CO 
CM 



00 
Oi 



O) O 



O) a> 
r^ CO 



0) o 

5 " 05 
« CO a. 

O 



oj c ra fl, o> 5 m- 
O O 



I 
-n <1> 0) 

.9? ra <D S" m ,„ <D tr 



,eS 




c 
o 

CO 

o 

O (0 

■>- n 

0) CQ 

-^co 



o 

cc 
CO 

Q. 
0) 

.> 
t> 

(0 

■~ ST 

•C CM 

c o 

1 li 
>s 

U M 
CO <0 

Is 

^8 
v| 

Q." 

"^ <S 
0) > 

u s 

O (0 
0) c 

:& a> 

u ^ 
^ c 
12 
.2* fl> 
'm !^ 

>»^ 



^^ 

0) > 

CO ^ 

S c 

0) 

£ E 

1.1 
o « 

1° 
i_ « 
o <» 

■*" 0) 



•1§ 

2 c 

CO — 

£. C 

o o 

CO O) 

!>. 



28 



TABLE 13 
DOC (1978) Summary of Proportional Increments to Explained Variance 



Independent 


Degrees of 


Increase 


F 


Significance . 


Variable 


Freedom 


In R^ 


Value 


Level: p 


Base: 










DOC/1000 (1974) 


1 


.81267 


1787.69 


.001 


Adjustment Variables* 


9 


.09377 


22.92 


.001 


(a) Occupancy (+) 










(b) Short Stay Beds (+) 










(c) Percent Aged (— ) 








j 


(d) Percent Days of Care 








1 


due to Medicare (-f ) 










Region 


3 


.00768 


5.63 


.001 


PSRO Review 


1 


.00112 


2.46 


N.S. 


Rate Review 


1 


.00109 


2.40 


N.S. 


BaseX PSRO Interaction 


1 


.00022 


.48 


N.S. 


Adjustment X PSRO Interaction 


9 


.00467 


1.14 


N.S. 


Region X PSRO Interaction 


3 


.00616 


4.52 


.01 


Rate Review X PSRO Interaction 


1 


.00034 


.75 


N.S. 


TOTAL 


29 


.92772 



Individual adjustment variables which were significant at the .05 confidence level are listed separately. The sign of the 
coefficient is in parentheses. 
N. S. = Not Significant 



PSRO review were included: (1) base; (2) other ad- 
justment variables; (3) region; and (4) rate review. 

With one exception, the interaction variables did not 
significantly affect the explanatory power of the equa- 
tion. The base rate by PSRO interaction, the set of ad- 
justment variables by PSRO interactions, and the PSRO 
by rate review interaction all had insignificant effects. 
However, the set of PSRO by region interaction terms 
was found to be significant at the .01 confidence level. 
What this indicates is that there are significant PSRO 
effects on DOC but that the effects differ among the 
four Census regions. Further, in light of the fact that 
the PSRO variable by itself indicated no effect for 
PSRO across the nation, it is likely that the PSRO 
impact in at least one of the regions is in the opposite 
direction from the others. Under this condition, the 
net effect of opposing PSRO impacts would be to at 
least partially negate each other. 

The results of the nationwide DOC regression anal- 
ysis strongly suggest that further analysis should be 
conducted at the Census region level of aggregation. 
The region by PSRO review interaction terms clearly 
indicate that PSRO i-mpact varies by Census region; 
regional regressions enable one to test whether PSROs 
have decreased (or increased) DOC relative to inactive 
PSRO areas within their region of the country. 

2.2.5.2 CENSUS REGION RESULTS: DOC 1978 

To determine whether the PSRO program had an im- 
pact on utilization within a region, a variant of the 
model developed to test for nationwide impact was 
used. The major difference between the two models is 



that the interaction terms were deleted. The interaction 
terms were deleted for two reasons. First, with the 
exception of the region by PSRO interaction, none of 
the interaction terms had a statistically significant 
effect. The importance of the region by PSRO inter- 
action is maintained by having four region-specific 
regressions. Second, at the regional level the interac- 
tion variables were highly collinear (correlated) with 
the regression's primary (non-interaction) variables. A 
generally accepted method of dealing with correlated 
sets of independent variables is to remove the offend- 
ing variables which make the least theoretical and 
interpretive sense. ^ 

In addition, there was a further respecification in 
the South region. In the South there is only one State, 
Maryland, which has a rate review commission. In 
addition, all of the PSRO areas in Maryland are active. 
Thus, the addition of a rate review variable in the South 
is tantamount to testing for a Maryland effect. There- 
fore, the rate review variable was deleted from the 
regression specification in the South. 

Norttieast Region: Table 14 presents the results of 
the regression equation estimated for the Northeast 
region. There are 46 PSRO areas in this region. As of 
July 1, 1978, 35 had begun review. The average time 
since beginning review for the 35 active areas was 
25.2 months. 

The results indicate that the base days of care ex- 
plained 75 percent of the variation in DOC. The set of 



' Intriligator, M. D., Econometric Models, Techniques and 
Applications. Prentice-Hall. Englewood Cliffs, New Jersey, 
1978. 



29. 



TABLE 14 

DOC (1978) Summat7 of Proportional Increments to Explained 
Variance: Northeast Region 



Independent 


Degrees of 


Increase 


F 


Significance 


Variable 


Freedom 


In R2 


Value 


Level: p 


Base: DOC (1974) 


1 


.75406 


236.60 


.001 


Adjustment Variables 


9 


.09183 


3.20 


.01 


(a) Change in Short Stay Bed 










Supply (+) 










PSRO Review 


1 


.03470 


10.89 


.001 


Rate Review 


1 


.01424 
.89483 


4.47 


.05 


Totals 


12 





adjustment variables were significant. Tal<en indi- 
vidually, however, only the change in short stay bed 
supply was significant. As expected ,an increase in the 
number of short stay beds per 1000 population was 
associated with an increase in DOC. 

The PSRO variable was significant and in the favor- 
able direction. PSRO review increased the explained 
variance by about 3.5 percent. The existence of rate 
review activity was also statistically significant: rate 
review is associated with an increase in DOC. 

The multiple regression results support the descrip- 
tive results presented in Table 7. The existence of 
PSRO concurrent review activity in the Northeast ap- 
pears to be associated with a reduction in DOC. 

North Central Region: Table 15 presents the sum- 
mary results from the regression analysis in the North 
Central region. There are 51 PSRO areas in this re- 
gion, of which 24 were active as of July 1, 1978. 



These 24 active areas had an average longevity of 
20.9 months. 

The results indicate that the base days of care in 
1974 accounted for 57 percent of the variance in 1978 
DOC. The set of adjustment variables increased the 
explained variance by an additional 30 percent. 

Taken individually, four adjustment variables had 
significant effects on days of care/1000 beneficiaries: 
occupancy, change in short stay bed supply, change 
in physician supply, and change in percent aged. The 
first three are associated with increases in DOC; the 
latter is associated with a decrease. 

The PSRO review variable is also statistically sig- 
nificant. PSRO review accounts for 1.4 percent of the 
variance in DOC; the sign of the coefficient indicates 
that the impact is favorable, i.e., reduces days of care. 

In the North Central Region, rate review did not have 
a significant effect on DOC. 



TABLE 15 

DOC (1978) Summary of Proportional Increments to Explained 
Variance: North Central Region 



Independent 
Variable 



Degrees 
Freedom 



Increase 
In R2 



F 
Value 



Significance 
Level: p 



Base: DOC (1974) 
Adjustment Variables 

(a) Occupancy (+) 

(b) Change in Short Stay Bed 
Supply (+) 

(c) Change in Physician 
Supply (+) 

(d) Change in Percent Aged (— ) 
PSRO Review 

Rate Review 
Totals 



12 



.57110 
.30032 



.01424 
.00761 
.89327 



203.34 
11.88 



.001 
.001 



5.07 
2.71 



.05 
N.S. 



30 



TABLE 16 

DOC (1978) Summary of Proportional Increments to Explained 
Variance: South Region 



Independent 
Variable 




Degrees of 
Freedom 


Increase 
In R2 


F 
Value 


Significance 
Level: p 


Base: DOC (1974) 
Adjustment Variables 

(a) Occupancy (-)-) 

(b) Change in Short Stay 
Supply (+) 

PSRO Review 
Totals 


Bed 


1 
9 

1 
11* 


.65837 
.13861 

.02636 
.82334 


167.70 
3.92 

6.71 


.001 
.001 

.01 



* In the South, as indicated earlier, the rate review variable was not included. This reduces the number of variables by one. 



South Region: The South region of the U. S. has 57 
PSRO areas. Of these, 23 areas, with an average 
longevity of 24.3 months, had begun review by July 1, 
1978. Table 16 presents the summary of results for 
the regression analysis in the South. The base DOC 
(1974) accounted for about 66 percent of the variance 
in DOC (1978). The set of adjustment variables in- 
creased the explained variance by an additional 14 
percent. Of the adjustment variables, two individual 
variables, occupancy and the change in short stay 
bed supply, had significant effects. Both of these vari- 
ables had positive coefficients indicating that increases 
in occupancy and in short stay bed supply are ac- 
companied tjy increases in the total days of care per 
1000 aged Medicare beneficiaries. 

Finally, the PSRO review variable increased the ex- 
plained variance by 2.6 percent. This was significant 
at the .01 confidence level. However, contrary to the 
results obtained in the Northeast and North Central 
regions, the sign of the coefficient is positive, indi- 
cating that the impact of PSRO review activities in the 
South is to increase days of care for Medicare 
beneficiaries. 

West Region: The West region has the smallest 
number of PSROs (35); with the exception of Cali- 



fornia, most of the western States contain only one 
Statewide PSRO. The West also has the highest rate 
of implementation among its PSRO areas (26 out of 
35). The average longevity of active PSROs (28.7 
months) is longer in the West than in other regions. 

Table 17 shows the results of the West Census 
region regression analysis. The 1974 base DOC ac- 
counted for 81 percent of the variance. The set of ad- 
justment variables was significant, reducing the ex- 
plained variance by 9.6 percent. Only one adjustment 
variable, Medicare-certified long term care beds, was 
statistically significant. The effect of this variable was 
to increase the days of care rate. That is, increases in 
the supply of these long term care beds were asso- 
ciated with increases in short stay days of care rate. 
This is a somewhat anomalous finding. One would 
expect that, as an alternative to acute care level, the 
availability of long term care beds would serve to 
decrease short stay hospital utilization. However, in the 
West region, there may not exist a backlog of Medi- 
care beneficiaries waiting to be discharged to long 
term care facilities. First, the days of care per 1000 
aged Medicare beneficiaries is much lower in the West 
than in other regions (26% lower than the next lowest 
region, the South). Second, despite the already low 
rate, the West had the greatest decrease in days of 



TABLE 17 

DOC (1978) Summary of Proportional Increments to Explained 
Variance: West Region 



Independent 


Degrees of 


Increase 


F 


Significance 


Variable 


Freedom 


In R2 


Value 


Level: p 


Base: DOC (1974) 


1 


.81016 


207.42 


.001 


Adjustment Variables 


9 


.09608 


2.73 


.01 


(a) Medicare-Certified 










Long Stay Beds (+) 










PSRO Review 


1 


.00135 


.35 


N.S. 


Rate Review 


1 


.00649 


1.66 


N.S. 


Totals 


12 


.91408 







31 



care (active and inactive areas combined): 4.2 percent 
from 1974 to 1978. Third, the West has the highest 
number of IVIedicare-certified long term care beds of 
any of the four regions (30/1000 beneficiaries). Fourth, 
occupancy rates are lowest in the West region. All 
these facts are indicative of a situation in which the 
need to transfer patients from short stay to long stay 
facilities may not exist. This would explain why long 
stay bed availability would have no effect on short 
stay utilization; it does not explain, however, why the 
impact is seemingly positive." 

The PSRO review variable increased the explained 
variance by an insignificant 0.1 percent. The West is 
thus the only region in which the PSRO impact was 
nonsignificant, even though it was in a favorable 
direction. 

The above regional results indicate that the study's 
primary hypothesis is partially confirmed: PSRO con- 
current review does indeed appear to have an impact 



'Since the Social Security Act requires a tliree-consecutive- 
day iiospitai stay prior to admission to a skilled nursing 
facility (SNF) for program payment (42 USC 405 (a) and 
1395 X (i)), there exists a threshold or three-day minimum 
number of hospital days of care for all potential SNF 
transfers below which the DOC would not be expected to 
be reduced. Further, this requirement may in fact increase 
DISC and ALOS, because patients who may otherwise 
not be admitted to the hospital will be hospitalized to 
fulfill the three-day stay requirement. 



on DOC. However, one has to disaggregate the data by 
region to observe how this influence operates. 

2.2.5.3 CENSUS REGION RESULTS: DISC AND 
ALOS 1978 

Given the significant interaction between the PSRO 
and region variables and the lack of significant PSRO 
impact at the national level, analyses of the two re- 
maining outcome variables (DISC and ALOS) are re- 
ported by Census region only. 

As mentioned in Section 2.1.1, PSROs have two 
mechanisms with which to control use rates: preadmis- 
sion review and continued stay review. If admission 
review is effective in reducing unnecessary admissions, 
the obvious impact is a reduction in discharge rate.*^ 
If continued stay review is effective in preventing 
excessively long stays, there should be a consequent 
decline in average length of stay. In addition, PSRO 
review activity may generate a "patrolman on the beat" 
deterrent effect where admissions or ALOS are re- 
duced simply because the PSRO mechanism exists. 

Model specifications for the DISC and ALOS anal- 
yses were essentially the same as for the DOC anal- 

" It is possible that admission review could affect ALOS as 
well. Admission review could reduce the number of very 
short stays (one or two days) by shifting treatment to an 
outpatient setting. This would in turn cause ALOS to 
increase. 



TABLE 18 

Summary of Proportional Increments to Explained Variance for 
DISC and ALOS Regressions: Northeast 

Discharges/1000 



Independent 
Variable 



Degrees of 
Freedom 



Increase 
In R2 



F Value 



Significance 
Level: p 



Base: DISC (1974) 
Adjustment Variables* 

(a) Change in Percent Aged (+) 
PSRO Review 
Rate Review 
Total 



12 



.90081 
.01774 

.00416 
■01211 
.93482 



456.07 
1.00 

2.11 
6.13 



.001 
N.S. 

M.S. 
.05 



ALOS 



Independent 
Variable 


Degrees of 
Freedom 


Increase 
In R2 


F Value 


Significance 
Level: p 


Base: ALOS (1974) 
Adjustment Variables* 

(a) Medicare-Certified Long Stay 
Beds (— ) 
PSRO Review 
Rate Review 
Total 


1 
9 

1 

1 

12 


.80050 
.06831 

.01600 
.03174 
.91655 


316.55 
3.00 

6.33 
12.55 


.001 
N.S. 

.05 
.001 



Individual adjustment variables which were significant at the .05 confidence level are listed separately. The sign of the 
coefficient is in parentheses. 
N. S. = non-significant. 



32 



yses. Regressions were run separately for each region 
and the adjustment variables remained the same as in 
the DOC analysis. For the DISC analysis, the outcome 
variable was DISC 1978, and the base rate was DISC 
1974. For the ALOS analysis, the outcome variable was 
the ALOS 1978 for all discharges involving aged Medi- 
care beneficiaries; the base rate was the correspond- 
ing ALOS in 1974. 

Findings 

Northeast: Table 18 presents a summary of incre- 
ments to explained variance for the discharge rate and 
ALOS in the Northeast. Whereas the days of care anal- 
ysis indicated a strong PSRO impact in reducing days 
of care, the discharge rate regression shows that the 
PSRO review variable failed to have a significant im- 
pact in increasing the explained variance (F=2.11). 
However, the sign of the regression coefficient indi- 
cates that the effect, although nonsignificant, was in 
the direction of decreasing discharges. In the ALOS 
regression, the PSRO review variable increased the 
explained variance by 1.6 percent, and this was sta- 
tistically significant (p<.05). Thus, it appears that the 
days of care savings in the Northeast region were 



more heavily influenced by continued stay review 
(ALOS) than by admission review (DISC). This is the 
first instance of ALOS influencing DOC. Previous 
evaluations have found DISC as the variable most 
likely to influence DOC. Similarly the Medicaid Rate 
Analysis (see Section 2.4) suggests that DOC is pri- 
marily influenced by DISC. 

North Central: Table 19 presents the discharges/ 
1000 and ALOS results for the North Central region. 
Whereas the days of care regression indicated a de- 
crease in days of care due to PSRO review, the dis- 
charge rate and ALOS analyses fail to disclose a sig- 
nificant PSRO impact on either measure (F=2.73 in 
the discharge rate regression and F^.51 in the ALOS 
regression). However, the regression coefficients for 
PSRO review in both equations are negative, indicating 
a favorable PSRO impact on both measures. Thus, al- 
though the impact of PSRO review is insignificant on 
both ALOS and the discharge rate, the gross effect is 
to reduce the days of care rate by a significant 
amount. This analysis, however, does not indicate 
whether admission or continued stay review is the 
more effective PSRO mechanism in accomplishing 
that reduction. 



TABLE 19 

Summary of Proportional Increments to Explained Variance for 
DISC and ALOS Regressions: North Central 

Discharges/1000 



Independent 
Variable 



Degrees of 
Freedom 



Increase 
In R2 



F Value 



Significance 
Level: p 



Base: DISC (1974) 


1 


.88061 


515.21 


.001 


Adjustment Variables* 


9 


.04878 


3.17 


.01 


(a) Occupancy (+) 










(b) Change in Short Stay Bed 










Supply (+) 










(c) Change in Percent Aged (— ) 










PSRO Review 


1 


.00467 


2.73 


N.S. 


Rate Review 


1 


.00099 


1.58 


N.S. 


Total 


12 


.93505 







ALOS 



Independent 
Variable 



Degrees of 
Freedom 



increase 
In R2 



F Value 



Significance 
Level: p 



Base: ALOS (1974) 


1 


.89862 


241.28 


.001 


Adjustment Variables* 


9 


.05876 


6.11 


.001 


(a) Medicare-Certified Long Stay 










Beds (— ) 










PSRO Review 


1 


.00055 


.51 


N.S. 


Rate Review 


1 


.00148 


1.39 


N.S. 


Total 


12 


.95941 







* Individual adjustment variables which were significant at the .05 confidence level are listed separately. The sign of the 
coefficient is in parentheses. 
N. S. = non-significant. 



33 



South: Table 20 presents the discharge rate and 
ALOS regression results for the South region. The 
PSRO review variable increased the explained vari- 
ance in the discharge rate regression and the ALOS 
regression by 2.1 percent and 3.7 percent, respectively. 
In neither case, however, is the effect statistically 
significant at the .05 confidence level. However, both 
are significant at the .10 confidence level. Also, both 
coefficients are positive, indicating that the PSRO 
impact, in the South, is to increase both the discharge 
rate and the ALOS. 

West: Finally, Table 21 shows the regression results 
for the discharge rate and the ALOS in the West re- 
gion. As previously noted, PSRO review does not have 
a statistically significant impact on days of care in this 



region. Not surprisingly, there is little indication of a 
PSRO impact on either discharges or ALOS. 

Summary: The results of the DISC and ALOS anal- 
yses provide a modest enhancement to the results of 
the DOC analysis. With the exception of the Northeast 
region, there is no statistically significant evidence of 
a differential impact of PSRO review on either ALOS 
or discharge rates. In the Northeast, where the PSRO 
impact on DOC is strongest, it appears that the DOC 
impact is due largely to a decrease in ALOS. Al- 
though the methodology does not enable one to dis- 
tinguish between admission review effects and con- 
tinued stay review effects, the results suggest that, at 
least for PSROs in the Northeast, continued stay re- 
view appears to be the more important mechanism in 
reducing overall hospital utilization. 



TABLE 20 

Summary of Proportional Increments to Explained Variance for 
DISC and ALOS Regressions: South 

Discharges/1000 



Independent 
Variable 


Degrees of 
Freedom 


Increase 
In R2 


F Value 


Significance 
Level: p 


Base: DISC (1974) 
Adjustment Variables* 

(a) Poverty (+) 

(b) Change in Short Stay Bed 
Supply (+) 

(c) Change in Physician 
Supply (-) 

PSRO Review 
Total 


1 .94425 
9 .02072 

1 .00213 
11 .96711 

ALOS 


1356.86 
3.31 

3.06 


.001 
.005 

N.S. 


Independent 
Variable 


Degrees of 
Freedom 


Increase 
In R2 


F Value 


Significance 
Level: p 


Base: ALOS (1974) 
Adjustment Variables* 
PSRO Review 
Total 


1 

9 

1 

11 


.91663 
.02798 
.00365 
.94826 


933.60 
3.17 
3.72 


.001 
.005 
N.S. 



* Individual adjustment variables which were significant at the .05 confidence level are listed separately. The sign of the 
coefficient is in parentheses. 
N. S. = non-significant. 



34 



TABLE 21 

Summary of Proportional Increments to Explained Variance for 
DISC and ALOS Regressions: West 

Discharges/ 1000 



independent 


Degrees of 


Increase 


F Value 


Significance 


Variable 


Freedom 


In R2 


Level: p 


Base: DISC (1974) 


1 


.81036 


152.77 


.001 ii 


Adjustment Variables* 


9 


.07175 


1.50 


N.S. 


(a) Medicare-Certified Long Stay 










Beds (+) 










PSRO Review 


1 


.00062 


.12 


N.S. 


Rate Review 


1 


.00057 


.11 


N.S. 


Total 


12 


.88330 








ALOS 






Independent 


Degrees of 


Increase 


F Value 


Significance 


Variable 


Freedom 


In R2 


Level: p 


Base: ALOS (1974) 


1 


.80030 


179.59 


.001 


Adjustment Variables* 


9 


.08074 


2.01 


N.S. 


(a) Medicare Certified Long Stay 










Beds (+) 










(b) Change in Percent Aged (-|-) 










(c) Percent Days Due to 










Medicare (— ) 










PSRO Review 


1 


.00018 


.04 


N.S. 


Rate Review 


1 


.02073 


4.65 


.05 


Total 


12 


.90196 







* Individual adjustment variables which were significant at the .05 confidence level are listed separately. The sign of the 
coefficient is in parentheses. 
N. S. = non-significant. 



2.2.6 Results: Impact Analysis 

Given that PSRO review has been found to have a 
statistically significant effect on Medicare days of care 
rates, it is appropriate to estimate the magnitude of the 
PSRO impact on DOC in order that dollar values for 
PSRO benefits may be estimated. In this section, esti- 
mates of PSRO impact in terms of days of care re- 
duced (increased) are calculated. For three of the 
Census regions, these estimates are statistically ap- 
propriate since PSRO review was found in those re- 
gions to have a significant impact on DOC. However, 
in the West region, the PSRO coefficient did not attain 
statistical significance. Thus, an estimate of impact 
would be based on a coefficient which regression has 
not shown to be different from zero. A possibility, 
therefore, would be to exclude the West from the 
analysis. However, because the PSRO program is a 
national program, it does not seem appropriate from 
a programmatic viewpoint to estimate the impact of 
the program for only certain parts of the country. An 
alternative is to estimate the PSRO impact in the West 
to be zero. Yet, although the coefficient does not differ 
from zero in a statistical sense, it is still the best point 



estimate of PSRO impact. Therefore, the impact esti- 
mates will include an estimated impact in the West 
region. The West impact, however, is accompanied by 
the caveat that it is based on an insignificant 
coefficient. 

The estimate of reduction in DOC due to PSRO re- 
view involves the calculation of two expected days of 
care rates. These "expected" rates are calculated by 
applying the average values of the base and adjust- 
ment variables for the active PSROs to the regression 
coefficients produced by the analysis. The first ex- 
pected rate uses the average value of the base rate 
and adjustment variables and sets longevity equal to 
zero. This gives an expected days of care rate for ac- 
tive PSRO areas, under the assumption that those 
areas had not become active (i.e., longevity=0). The 
second expected days of care rate uses the same re- 
gression coefficients and the same average values for 
the base and adjustment variables, but this time sets 
longevity equal to its average value. This gives an 
expected days of care rate for active PSRO areas, 
given the average length of time that those areas had 
been performing review functions. 



35 



The difference between the first expected rate (no 
PSRO in operation) and the second (PSRO in opera- 
tion) is a measure of PSRO impact. By dividing this 
difference by the first rate (and multiplying by 100), 
one obtains the estimated percent reduction in DOC 
due to PSRO review. This can be expressed as follows: 



Percent days saved = 



Exp (w/o) — Exp (w) 



X 100 
Exp (w/o) 

where: Exp (w/o) = expected days of care/1000 
without PSRO 

Exp (w) = expected days of care/ 1000 
with PSRO 

Table 22 presents the unstandardized regression co- 
efficients used in the impact analysis. These coeffi- 
cients are taken from the four regional regressions re- 
ported in Section 2.2.5.2. 

The results of the impact estimation are presented 
in Table 23. The first two columns in the table are the 
estimated days of care rates with and without PSRO 
activity. The third column is the difference between the 
two and is the estimate of PSRO impact on DOC (a 
negative value indicates that the PSRO impact was to 
decrease the days of care rate). Thus, in the North- 
east, North Central, and West regions, the estimated 
effect of PSRO review was to reduce days of care/ 
1000 by 203, 87 and 39, respectively, whereas in the 
South, the PSRO impact was to increase days of care/ 
1000 by 135. As indicated in column (4), the percent 
change in days of care ranged from the 4.8 percent 
reduction in the Northeast to the 3.7 percent increase 
in the South. Estimated Days Saved is obtained by 
multiplying the PSRO percentage impact by the Medi- 
care aged enrollment for active PSROs and dividing the 
result by 1000 (because the percent days saved are 
in rate per 1000 form). The results of these calcula- 
tions are presented in Column 6. The estimate of days 



saved for the four regions combined is approximately 
950,000 days. This is approximately the number of days 
saved in the Northeast region alone. In effect, then, 
the number of days saved in the West and North 
Central regions is counter-balanced by the increase in 
days of care in the South., 

A national aggregate estimate of percent of days 
saved was calculated by dividing the total days saved 
(948,430) by an estimate of the number of days that 
would have been used by Medicare aged beneficiaries 
in the active PSRO areas had the PSROs not been 
active (56,320,806). Consequently, the estimate of the 
national aggregate impact of the PSRO program on 
Medicare days of care is a reduction in days of care 
by 1.7 percent. 

2.2.7 Discussion of Results 

In summary, the results of the 1979 Medicare Impact 
Study (100 Percent Medicare Claims File study) indi- 
cate an overall PSRO impact of approximately 1.7 
percent reduction in days of care/1,000 aged Medi- 
care beneficiaries. This finding is consistent with the 
results of the 1978 100 Percent Medicare Claims File 
study. In the 1978 model specification which most 
closely approximates this year's specification, an over- 
all reduction in DOC of 2 percent was found. By re- 
gion, the impact estimates from the two studies are 
fairly consistent as shown in Table 24. The Northeast 
and North Central regions show slightly more favorable 
impacts in this year's analysis as compared to last 
year's. The South, which showed an unfavorable im- 
pact of PSRO review in the 1978 study, shows an un- 
favorable impact again in this year's study. Finally, the 
West region, which showed a favorable PRSO impact 
in the 1978 analysis, fails to show a significant impact 
in this year's analysis; however, the direction of the 
impact is toward a reduction in DOC. 



36 



ra 
O 



a 
o 



o 
S 



V 

u £ 
m J2 



.2 

M 
M 
0> 

k. 

bO 
a> 

■o 

V 
N 



c 



c 
a> 
'o 

3= 
0) 
O 

o 

c 
_o 
'in 

V) 

<o 

O) 
0) 

k. 

T3 
<U 
N 



C 

o ^^ 

0) CO 

k. II 

0) ^-• 



g 

or: 
2 In 



^5. 



05^.^^-»C0. — . CM . — .^i^ IT) 

■inor^in cm t-co o 



r^ o in T- o 
(35 o O) m cm 
I- 1- OT in 1- 

O Tt CM CM •* 
O) <o o o T- 

■ cj T-' ■ o 

CM I 1- 



U5 o ^^^^in 

(O O Oi ■^ (D 



o 

00 

00 

CM 
CO 

CO 



00 CM 



<j) r^ 



CM 






c 
o 
o> 

Q> — 
«- CO 

*- ■^ 

IS II 

OZ 



00 U) O) CO CM 

CO CJ) 00 1- in 

CO CO 1- o •^ 

CO •^ CM T- in 

CO r- o> o o 

' cvi ■ "cm 

CO CM 



00 ^-. 

CO CO ^^^^ 
■ 1- O CO 



•I- in 



^ CM 



00 O O CO o 

O CO CM r~- CM 

C3) CM o in -i- 

CM CO ^ O O 

C3> in '^ C3 O) 

CM ' f CJJ 

oi m 

CM 



CO . — . CO . — , 00 
0> ^ CM CO 05 

1— O CO O) 00 

00 1- T- T- CO 

in O) i^ CO in 
r^ r^ o C3) 
r^ C3) o t^ 



o 

CM 

o 
00 

00 

CM 



CJ) 



CO 
CO 



CO 



CO T- 

in I 



CO 
CO 



o 
o 
o 



(0 

c 
o 
« 

(U 
CI 

o 
o 
o 



o 
o 
o 



Si 






0) 

CO o 
DO 



CO *- -r 

CO Jr C -I— 

"^ .2 - I 
"H— CJir^ 

D C 0> 
> Q- CO J3, 
O O JC 
Q. D. O 



o 
o 
o 



0) 
■D 



>. 
CO 



D) 

c 

0) 



O O 

in CM 
00 t^ 
CO in 
00 in 
r^: 00 



CM 
CD 
CO 

in 



00 
CM 



o 

00 
CO 

in 
in 



^^CJ) 
^ CM 
00 • 



p O O) 

r~- Tj- -,- 
m o CO 
CO in CO 

CO ^ O) 

T-' CT) -r-" 

I •^ CO 



^-^ 


r^ -~. 


^ 


o 


CJ) 




CO 


CM 00 


t^ 


■<3- 


in 




CO 


^' 1^ 


s. 




2J, 




o 


vr. ° 


CM 


00 


T— « 


ro 


CNJ 


CM O 


o 




o 


Tt 


T— 


CJ) O 


in 


■<;l- 


CO 


(T) 


00 


CM in 


o 


00 


CO 


CO 


in 


00 CO 


o 


■<3- 


in 


CM 



00 r^ 

O) I 



CM 



in 



in O) 

O CO 



m 
o 

CM CM T-^ 



CO 



o o 

CM CM 
<3) CO 

in CD 
CO •^ 

■1- CO 
O I 



co° 

CO • 



CD CM 

in CM 

O CO 

in o 
q q 
d CO 

O 1- 

CM I 



O 

or 



<3) in 

o o 

CD tT 

O 1- 

00 m 



in 

CJ) 



m 

CM 
5 



00 



CJ) 

in 

CO 
00 
00 



CO in 

CO CO 

CM f-^ 



O O CO 

CO r^ 00 

CM t^ CJ) 

in CJ) CO 

1- q IV- 

Tf r-^ CD 

I O) o 

I I CO 



CO 1- 
CM T- 

CO CM 



CM r^ CJ) 

CO CD CM 
""d- "* CJ) 
CD CM CO 

o 1- q 
ocj iri CM 

I (3) O 
I T- CD 



Si 

Q. 

-^ CO 
CO H- 
C O 

.5 ^ 

o o> 



(/) 

>. 

(0 
■D 

O 

O 

•O 



O 
O 



Q) 



•55 -g 



CJ) o 



CO 



Q. C 



CO 0) 
O C 

O) •" 



c c «-: 



■-^ </) 

<- >> ' 

.2 TJ 



in 

T3 
C 0) 

.2 -"^ 
i ra CO 

^ « TJ 



o r: J2 

fc- ^ -.— _ •— to 
- ■- (0 P -g D) 1 



.2 

"> 

0) 

o 

C 



1.2 

o II 

CO C 



0) 

O) en 
c c 
CO CO 
.c .c 



3 CO 

D. O 



0^-^ 





I- 



^lo 



0.2*- 
« 2 « 

O (0 o 



CC o 



^ .> 

O O 

■" Q) 

« S) 

0) a> 3 

(0 o o 

c c/3 



CO o c 

Q. o — 

»- c: 

c o .2 



0) (1) 

to a) 



CO 
3 
CT 

a> 

c 
o 



O CO 

;^ "O (D 

Q) CO ^ 

O 0) 

O T- >- 

O 0} 

I si 

CD P 

;: 0) ^ 

<D ■*- ^ 
>- <- 0) 

.«| 

o 



CO 

^ *^ CO 

_2 CO fl) 

CO Q) ■> 

> >- o 

J!- CO w 

Q) O^^S 

^ c\i -Jo 

H- CO DC 



37 



TABLE 23 
Estimated PSRO Impact on Days of Care per 1000 Aged Medicare Beneficiaries, by Region 



Region 



(1) 

Estimated 

Days of 

Care/ 1000 



(2) 



(3) 



Estimated PSRO Impact (4) 

tstimatea ^^ pgpQ percent 

Without PSRO With PSRO ^fsT-lT ^^^^^^^ 



(5) 

Medicare 

Aged 

Enrollment* 



(6) 

Estimated 

Days Saved 

(3)X(5)-^1000** 



Northeast 
North Central 
South 
West*** 



4240 
4106 
3691 
2787 



4037 
4020 
3826 
2748 



—203 
—87 

+135 
—39 



—4.8% 
—2.1% 
+3.7% 
—1.4% 



4,728,898 
3,783,674 
3,383,325 
2,958,974 



—959,924 
—328,612 
+456,039 
—115,933 



Nation 



3791 



3728 



-64 



-1.7% 



14,854,871 



—948,430 



Enrollment is for active PSRO areas only. 

Column (6) does not equal (3) X (5) -^ 1000 exactly due to rounding of column (3). 

The estimate of PSRO impact in the West is based on an insignificant regression coefficient. 



TABLE 24 

Estimated Percent Reduction in DOC: 1978 and 1979 
Study Results 





Percent Reduction 


Region 


1978 Study 
Results 


1979 Study 
Results 


Northeast 
North Central 
South 
West 


-4.0% 
—1.6% 
+1.3% 
—2.0% 
-2.0% 


—4.8% 
—2.1 % 
+3.7% 
—1.4%* 


Total 


-1.7% 



* Based on an insignificant regression coefficient 



The results from the two years' studies are not strictly 
comparable. The PSRO program has had an additional 
year of maturity between the two studies. Thus, some 
of the PSROs which were inactive in last year's study 
are active in this year's. This will change impact esti- 
mates. Some changes were made in the model spe- 
cification with regard to the adjustment variables. In 
particular, the measure of long term care bed supply 
was changed in this year's analysis. An additional 
variable, percent of short-stay beds located in teaching 
facilities, was added to this year's analysis. These 
differences will also change the impact estimates 
slightly. 

Another difference from the 1978 100 Percent Medi- 
care Claims File study is in the selection of hospitals 
to be included in the analysis. Additional edits were 
made in this year's study to eliminate Federal hospitals 
and hospitals which, although coded as short-stay 
facilities, actually are more similar to long term care 
facilities. These edits resulted in 209 hospitals being 
deleted from this year's analysis. These hospitals rep- 
resented 17,000 Medicare discharges and 630,000 
Medicare days of care In 1978. 



In assessing the results of this year's Medicare Im- 
pact Study, certain limitations should be kept in mind. 
First, the implementation of PSRO review was not a 
random process. PSROs self-selected themselves 
into the active review category. This could have a 
bearing on the results. However, in the absence of a 
random design, one way to control for such a bias 
is the use of adjustment variables in the forced order 
regression model. ''^ 

Second, the study estimates PSRO impact on Medi- 
care utilization only. It does not address the issue of 
whether PSROs are more or less effective with respect 
to Medicaid and Maternal and Child Health utilization. 

Third, the only measure of PSRO review activity 
used this year is length of time of review. It would be 
valuable to know, as well, the operational character- 
istics of PSROs which permit one to differentiate 
between successful and unsuccessful PSROs. At this 
time, measures of organizational effectiveness for 
PSROs are in the developmental stage (see Section 7). 

Fourth, in this analysis active PSRO review was 
compared with inactive PSRO review. While it is not 
clear exactly what review activities are taking place 
in the inactive areas, some review is undoubtedly be- 
ing done in those areas. Therefore, the PSRO impact 
reported in this study is the incremental impact of 
PSRO review over other forms of utilization review, 
rather than the impact of PSRO review over a total 
absence of review. 

Finally, the finding of such great differences in 
measures of PSRO effectivenesses by region raises a 
host of questions which cannot be answered without 
further research. The PSRO program is a national pro- 

" Another method of controlling for a self-selection bias 
would be to treat implementations of PSRO review as an 
endogenous variable in a set of simultaneous equations. 
One could then partial out the effects of regional, 
demographic and economic characteristics in the selection 
process. This technique is being explored for future 
analysis. 



38 



gram in that the basic goals of improving efficiency 
and maintaining quality exist for all PSROs in all 
regions. Therefore, one would expect no statistically 
significant geographic differences in PSRO effective- 
ness unless the regional variable serves as a proxy 
for some factor or factors which do affect PSRO per- 
formance. There are as yet no explanations for these 
findings, nor is there any assurance that the four 
Census regions are the most appropriate geographic 
units. One can only state with assurance that this is 
a promising area for investigation. 



2.3 PSRO Impact on Specific Diagnoses 
and Procedure Categories Study 
(Diagnosis- and Procedure-Specific 
Impact Study) 

2.3.1 Background 

The 1978 PSRO evaluation showed that, relative to 
inactive PSRO areas, active PSROs reduced DOC by 
a statistically significant 1.5 percent. Given this result, 
this Diagnosis- and Procedure-Specific Impact Study 
examines utilization aspects of four specific surgical 
procedures and one diagnosis which are hypothesized 
to be affected by PSRO activity. In addition, several 
procedures and diagnoses are chosen for which no 
PSRO impact is expected. Section 2.3.2 presents a 
literature review which develops the hypotheses for 
the procedures and diagnosis selected for examina- 
tion. Section 2.3.3 shows the methods used in this 
study, describing both the analysis plan and the data 
base used. The final section, 2.3.4, presents descriptive 
and analytical findings, both in terms of the individual 
procedures and diagnosis, and the conclusions which 
can be drawn from these results. 

Before commencing with the literature review, sev- 
eral limitations of this study should be discussed. In 
a study such as this, the selection of procedures and 
diagnoses is a key. For this study, conditions were 
selected via a literature review, combined with con- 
sultation with those knowledgeable in the field. Thus, 
although several of the conditions were mentioned in 
the literature of the National Professional Standards 
Review Council and American Association of PSROs, 
this does not imply that active PSROs necessarily gave 
them top priority. Therefore, the selection of proce- 
dures and diagnoses may be subject to interpretation. 
In addition, the actual data used for analysis may 
be subject to some coding irregularities, it appears 
that a large number of procedures and diagnoses are 
not consistently coded into the Medicare records," and 
this may impact on this study. The issue of data coding 
is discussed more fully in Section 2.3.3.2, Data Sources. 
One must be aware of these considerations when in- 
terpreting the following report. 



" Institute of Medicine, "Reliability of Medicare Hospital 
Discharge Records" (Washington, D.C., National Academy 
of Sciences, November 1977). 



2.3.2 Literature Review as Basis for Procedure and 
Diagnosis Selection 

2.3.2.1 INTRODUCTION 

In the past two decades, the costs and benefits of 
many health care activities have been evaluated. These 
evaluations have suggested that certain established 
practices are not only more expensive than available 
alternatives, but also are no more beneficial to the 
patient. For such care activities, it is reasonable to 
expect that the PSRO program would attempt to pro- 
mote a shift from the more expensive to the less ex- 
pensive approach. Evaluation of the PSRO program, 
therefore, should include analysis of the extent to 
which PSROs shift the style of practice toward these 
less expensive approaches to problems. 

In this section, five patient problems are selected for 
examination: one diagnosis (myocardial infarction), and 
four procedures (cholecystectomy, cataract removal, 
hysterectomy, and surgery for breast cancer). First the 
rationale for the selection of the diagnosis is presented. 
This is followed with the rationale for the selection of 
the four procedures. 

2.3.2.2 ACUTE MYOCARDIAL INFARCTION 

Within the last 15 years, the established treat- 
ment for acute myocardial infarction (AMI) has con- 
sisted of aggressive treatment regimens provided in 
highly specialized care units of a hospital followed by 
monitoring and bed rest, often lasting three to four 
weeks. Within the last 10 years, however, several re- 
ports have suggested that "low risk" patients could be 
discharged in far less time, thus reducing average 
length of stay without increasing patient risk. 

Several studies have randomly divided appropri^e 
patients into short-stay and longer-stay to evaluatelhe 
effect of ALOS on patient outcome. Hutter et al." stud- 
ied 138 patients with uncomplicated AMI, who were 
randomly assigned to 14 and 21 day hospitalizations. A 
24 week follow-up with respect to mortality, return to 
work, and rehospitalization indicated no significant 
differences in outcome due to ALOS. Thus, Hutter et al. 
argue that those patients with uncomplicated AMI re- 
ceived no additional benefits from a prolonged hospital 
stay. In a 1974 study, Hayes et al.'= followed 189 pa- 
tients with uncomplicated AMI who were randomly 
assigned to either discharge at nine days or dis- 
charge at 16 days groups. A six-week follow-up indi- 
cated no significant difference in mortality or morbidity. 
More recently, a 1976 study by Mather et al.^^ pre- 

^' A. M. Hutter, V. W. Sidel, K. I. Shine, et al., "Early hospital 
discharge after myocardial infarction." New England 
Journal of Medicine (N. Engl. J. Med) 228: 1141-1144, 1973. 

'°M. J. Hayes, G. K. Morris, and J. R. Hampton, "Comparison 
of mobilization after two and nine days in uncomplicated 
and myocardial infarction." British Medical Journal (Br. 
Med J) 3: 10-13, 1974. 

'" H. G. Mather, D. C. Morgan, N. G. Parson, et al., 
"Myocardial infarction: A comparison between home and 
hospital care for patients." Br. Med J: 925-929, 1976. 



39 



sented a more dramatic case, in that 450 patients with 
uncomplicated AMI were assigned to either hospital 
care or home care by the family doctor. Follow-up at 
both six weeks and one year indicated that those in 
the home care group had a slightly lower mortality 
rate than the hospital care group. 

As a result of these studies and others,*^ it is felt 
that many "low risk" AMI patients can be discharged 
from hospital in less time than was previously con- 
sidered the norm, 15.6 days.^^ McNeer, et al." report 
that 45 percent of their patients were eligible for 
early discharge. Since the number of patients with 
AMI in 1973 was 400,000,=" this would indicate that 
over 175,000 are eligible for early discharge. Thus, a 
reduction in average length of stay for uncomplicated 
cases could result in a substantial reduction in days 
of hospital care. 

2.3.2.3 SELECTION OF PROCEDURES 

The selection of procedures for inclusion in this 
study is essentially based on the premise that some 
unnecessary surgery exists. This section points to 
evidence suggesting unnecessary surgery and indi- 



" G. R. Royston, "Short stay hospital treatment and rapid 
rehabilitation of cases of myocardial infarction in a district 
hospital." Br Heart J 34: 526-535, 1972. 
H. H. Tucker, P. H. M. Carson, N. M. Bass, et al., "Results 
of early mobilization and discharge after myocardial 
infarction." Br Med J 1: 10-13. 1973. 
G. Rose, "Early mobilization and discharge after myocardial 
infarction." Modern Concepts Cardiovasculat Disease 41 : 
59-63, 1972. 

A. Blosh, J. P. Maeder, J. C. Haissly, et al., "Early 
mobilization after myocardial infarction: a controlled study." 
American Journal of Cardiology 34: 152-157, 1974. 

J. E. Harpur, R. J. Kellet, W. T. Connor, et al., "Controlled 
trial of early mobilization and discharge from hospital in 
uncomplicated myocardial infarction." Lancet 2: 1331-1334, 
1971. 

B. M. Groden, A. Allison, and G. B. Shaw, "Management 
of myocardial infarction: the effect of early mobilization." 
Scottish Medical Journal 12: 435-439, 1967. 

D. M. Boyle, J. M. Barber, M. J. Walsh, et al., "Early 

mobilization and discharge of patients with acute myocardial 

infarction." Lancet 2: 57-60. 

A. A. Adgey, "Prognosis after early discharge from hospital 

of patients with acute myocardial infarction.' Britisti Heart 

Journal 31: 750-752, 1969. 

P. Brummer, V. Kallio, and E. Tala, "Early ambulation in 

the treatment of myocardial infarction." Acta Med Scan 

180: 231-234, 1966. 

J. F. McNeer, A. G. Wallace, G. S. Wagner, et al., "The 

course of acute myocardial infarction: feasibility of early 

discharge of the uncomplicate patient." Circulation 51: 

410-412, 1975. 

" Length of Stay in PAS Hospitals, 1975. Ann Arbor, Michigan, 
Commission on Professional and Hospital Activities, 1976. 

"McNeer, J. F., Magner, G. S., Ginsburg, P. B., et al., 
"Hospital discharge one week after acute myocardial 
infarction." N Engl J Med 298: 222-232, 1978. 

™ N. C. Chaturvedi, M. J. Walsh, A. Evans, et al., "Selection 
of patients for early discharge after acute myocardial 
infarction." Br. Heart J 36: 533, 535, 1975. 



vidually examines the procedures selected for this 
study. 

Despite the limitation imposed by a lack of definition, 
discussions and debates on unnecessary surgery are 
taking place. Basically, there are four categories of 
studies that are used to demonstrate the existence 
of unnecessary surgery: 

• Variations in rates of surgery among different 
localities: A number of studies have found regional 
variations in utilization rates for a number of com- 
mon procedures.'^! In addition, correlations have 
been found between higher rates of surgery and 
the number of surgeons and facilities.^z Given that 
mortality rates do not differ among these regions, 
these data suggest that some medically unnec- 
essary surgery is taking place. 

• Variations in rates of surgery between fee-for- 
service and pre-paid delivery systems: Studies 
have shown that overall surgery rates, as well as 
those for specific procedures, may be twice as 
high for users of fee-for-service medical care as 
prepaid subscribers. '^ Since no distinct health dif- 
ference has been observed, this may indicate over- 
utilization in fee-for-service settings, underutili- 
zation in pre-paid delivery systems, or both. 

• Second Opinion for Surgery: This area is con- 
sidered somewhat controversial, due to the fact 
that it is difficult to say which of two dissenting 
opinions is the correct one. However, an ongoing 
study has found that 17.6 percent of the patients 

in a mandatory second opinion program who were 
recommended for surgery and required to have a 



■''^ P. A. Lembcke, "Comparative Study of Appendectomy 
Rates," American Journal of Public Health, 42: 276-286 
1952. 

M. Gornick, "Medicare Patients: Regional Differences in 
Length of Hospital Stays, 1969-71," Social Security 
Bulletin, July, 1975, pp. 16-33. 
C. E. Lewis, "Variations in the Incidence of Surgery," 
New England Journal of Medicine, 281: 880-884, Oct. 16, 
1969. 

J. Wennberg and A. Gittelsohn, "Small Area Variations in 
Health Care Delivery," Science, 182: 1102-1108, Dec. 14, 
1973. 

J. Wennberg and A. Gittelsohn, "Health Care Delivery in 
Maine: Patterns of Use of Common Surgical Procedures," 
Journal of the Maine Medical Association, 66: 123, May 
1975. 

"- J. P. Bunker, "A Comparison of Operations and Surgeons 
in the United States and in England and Wales," New 
England Journal of Medicine, 282: 134-144, January 15, 
1970. 

** C. R. Gaus, B. S. Cooper, and C. G. Hirschman, "Contrasts 
in HMO and Fee-for-Servlce Performance," Social Security 
Bulletin, May 1976, pp. 5-14. 

G. S. Perrott, The Federal Employees Health Benefits 
Program: Enrollment & Utilization of Health Services, 
1961-1968. DHEW publication no. (HSA) 75-13027. 
May 1971 (reprinted 1975). 



40 



second opinion were not confirmed for hospital 
surgery by the second physician. »* 
• Comparison of Reported and Approved Indications 
for Surgery: It has been found that following a 
review of indications for surgery, the proportion 
of unnecessary surgical procedures can be sig- 
nificantly reduced. =55 

The evidence regarding the reducibility of hospital 
utilization surgery was considered by the American 
Association of PSROs (AAPSRO) ad iioc Surgical 
Criteria Committee. The Committee, in a January 1978 
letter, reported 11 procedures "selected by our 
Committee as representing common procedures that 
we have learned through experience and communica- 
tions with PSROs across the country have a significant 
potential for inappropriate utilization." =6 This list is 
presented below. 
Procedures Selected by the AAPSRO ad hoc Surgi- 
cal Criteria Committee as Having a Significant 
Potential for Inappropriate Utilization: 

1. Dilation and Curettage 

2. Tonsillectomy/Adenoidectomy 

3. Cholecystectomy 

4. Inguinal Hernia Repair 

5. Lumbar Disc Excision 

6. Meniscectomy 

7. Appendectomy 

8. Vaginal Hysterectomy 

9. Abdominal Hysterectomy 

10. Coronary Arteriograph 

11. Cataract Removal 

From this list of 11 surgical procedures and 
from the review of medical literature, three procedures 
were selected for inclusion in this study: hysterectomy, 
cataract surgery, and cholecystectomy. In addition to 
these, it was felt that breast cancer surgery would also 
be an appropriate topic. ^ brief discussion of justifica- 
tion for each procedure, follows. 

2.3.2.4 HYSTERECTOMY 

Several studies have indicated the possibility of 
excess surgery occurring in this surgical area. A study 
by Dyck in Sasketchewan, as well as ones by Miller 



' E. G. McCarthy and G. W. Widmer, "Effects of Screening 
by Consultants on Recommended Elective, Surgical 
Procedures, New England Journal of Medicine, 291: 
1331-1335, December 19, 1974. 

F. J. Dyck et al., "Effects of surveillance on ttie number 
of liysterectomies in tlie province of Saskatchewan." 
N Engl J Med 296: 1326-1328, 1977. 

' N. F. Miller, "Hysterectomy: therapeutic necessity or 
surgical racket?" American Journal of Obstetrics and 
Gynecology (Am J Obstet Gynecol) 51: 804-810, 1946. 
J. C. Doyle, "Unnecessary hysterectomies: study of 6,248 
operations in thirty-five hospitals during 1948," Journal of 
ttie American Medical Association (JAMA) 151: 360-365, 
1953. 

'American Association of PSROs Ad Hoc Surgical Criteria 
Committee, John Bussman, M.D., Chairman. Letter to 
Members of the National Council of PSROs, January 8, 
1978. 



in the Midwest and Doyle in California all found a 
substantial percentage of unjustified hysterectomies 
following a review of indications for surgery.^^ A varia- 
tion in rates also exists for hysterectomy, as Wenn- 
berg and Gittleson =« found a range from 20 to 60 per 
10,000 population in a study in Vermont. These find- 
ings, coupled with the fact that the Dyck study showed 
the possibility of rate reduction, indicate that it is 
appropriate to look for PSRO effect of review on 
hysterectomy rates. 

2.3.2.5 CATARACT SURGERY 

Cataract surgery, a procedure common among the 
elderly population, has greatly increased in recent 
years. Whereas the population at risk has increased by 
27 percent and the number of ophthalmologists has 
increased by 25 percent from 1965 to 1977, the num- 
ber of cataract operations has increased by 177 per- 
cent over the same time period. ^^ This sharp rise is of 
concern due to the fact that "there are no firm criteria 
in existence which define the point at which a person's 
lens should be extracted. The formation of a cataract 
is usually an indolent process, and it is unlikely that 
there has been a sudden change in their frequency." ^° 
Of particular concern is the wide variation in rates of 
lens extraction. Although the rate for the United States 
as a whole is 9.9 per 1,000 Medicare enrollees, the 
range across PSRO areas extends from 5.1 to 16.8 per 
1,000 Medicare enrollees. •=! Therefore, there is a strong 
possibility that unnecessary surgery is occurring 
among cataract patients, and that active PSROs could 
have an impact by controlling the rising rates of 
cataract surgery. 

2.3.2.6 CHOLECYSTECOMY 

Cholecystectomy, a surgical procedure for removal 
of the gall bladder also common in elderly popula- 
tions, has also shown varied rates in different situa- 
tions. The study by Wennberg and Gittleson ^^ in Maine 
showed cholecystectomy rates ranging from 27 to 55 
per 10,000 population. In addition, Perrott^^ found that 
those members of the Federal Employees Health 
Benefits program enrolled in prepaid health plans 
showed a cholecystectomy rate of 1.5 per 1,000, 
whereas the rate for those obtaining fee-for-service 



'Dyck, op. cit.; Miller, op. cit.; Doyle, op. cit. 

'J. Wennberg and A. Gittleson, "Small Area Variations in 
Health Care Delivery", Science 182: 1102-08, Dec. 14, 1973. 

'J. Taylor, "Changes in the Rate of Cataract Extraction 
Since the Inception of Medicare," paper presented to 
American Public Health Association, New York, N.Y., 
Nov. 1979. 

'Taylor, ibid. 

^"Information on Surgical Procedure Rates to Aid in 
Objective Setting," General Memorandum No. 9-79, DHEW, 
HCFA, Health Standards and Quality Bureau. 

■J. Wennberg and A. Gittleson, "Health Care Delivery in 
Maine: Patterns of Use of Common Surgical Procedures," 
Journal of trie Maine Medical Association, 66-123, May 
1975. 

' Perrott, op. cit. 



41 



care was 2.1 per 1 ,000. Since this procedure is con- 
sidered hazardous for the elderly «* and there is the 
alternative of no therapy in some cases, it is advanta- 
geous to reduce any possible unjustified surgery. 

2.3.2.7 BREAST CANCER 

Breast cancer is an extremely common condition in 
the U. S., with some 100,000 newly diagnosed cases 
annually.^'^ For many of these patients, therapy is 
radical mastectomy, which is extremely risky, physi- 
cally and emotionally damaging, as well as costly. In 
addition, a series of studies have compared radical 
mastectomy to alternatives (especially modified radi- 
cal, simple mastectomy, and tylectomy) and found the 
alternatives to be equally effective. It has been sug- 
gested that less radical surgery might result in less 
damage to the patient, and a lower cost. 

In a 1965 study by Kaae and Johansen,^^ patients 
with newly diagnosed breast cancer were assigned at 
random to treatment with simple mastectomy includ- 
ing postoperative x-ray therapy or to radical mastec- 
tomy. The five- and ten-year survival rates for the two 
groups were nearly identical. After five years the 
survival rates for simple mastectomy and radical mas- 
tectomy were 57 and 58 percent, respectively. After 
ten years a slightly higher morbidity was reported for 
those with radical mastectomy. Utilizing 242 pa- 
tients ages 35 to 69 years between 1 964 and 1 971 , 
similar results were found regarding five-year sur- 
vival rates and freedom of disease in a study by 
Hamilton et al." 

A 1977 review concluded, "we have been unable to 
discover a single randomized series in which radical 
mastectomy, with or without radiotherapy to the axilla, 
has shown any benefit over simple mastectomy with 
radiotherapy." In addition, the differences "in terms of 
mutilation, morbidity, length of convalescence, and 
time length for the wound to heal are substantial." "* 

Breast cancer surgery is perhaps the most contro- 
versial of the diagnosis and procedures included in 
this analysis because the practice of performing other 
than radical mastectomies was just beginning to gain 
acceptance in 1977. However, malignant neoplasm of 



"G. Fitzpatrick, R. Neutra, and J. P. Gilbert, "Cost- 
effectiveness of cholecystectomy for silent gallstones," in 
Bunker, Barnes, and Mosteller, op. cit. 

•^W. Finkle, "Potentially Excessive Hospital Care," 
Unpublished Paper Nov. 1979, p. 15. 

"S. Kaae and H. Johansen, "Simple mastectomy plus 
postoperative irradiation by the method of IVIcWhirter for 
mammary carcinoma," Prog Clin Cancer 1: 453, 1965. 
S. Kaae and H. Johansen, "Simple versus radical 
mastectomy in primary breast cancer, Prognostic Factors 
in Breast Cancer," Edited by APM Forrest and P. B. Kunkler, 
Edinburg, E. S. Livingstone, 1968, p. 93. 

"'T. Hamilton, A. O. Langslands, and R. J. Prescott, "The 
treatment of operable cancer of the breast: a clinical trial 
in the south-east region of Scotland," Br. J Surg 61: 758, 
1974. 

"* K. McPherson and M. S. Fox, "Treatment of breast cancer," 
in Bunker, Barnes, and Mosteller, op. cit. 



the breast, along with the other diagnosis and proce- 
dures under study was one of the procedures approved 
by the National Professional Standards Review Coun- 
cil in 1975 as appropriate for PSRO evaluation.** The 
PSRO Technical Advisory Panel, which includes 
physician representatives of the National Professional 
Standards Review Council, felt it appropriate for in- 
clusion in this analysis. It was thought that the earliest 
active PSROs might be more medically innovative than 
inactive areas, and as a result would be in the fore- 
front of the movement away from the radical 
mastectomy. 

To summarize, the literature suggests that appro- 
priate reduction in utilization can be made without 
compromising quality in the following areas: 

Diagnosis or Procedure in which PSRO Impact is 
Expected 

1. Acute myocardial infarction 

2. Breast cancer surgery 

3. Hysterectomy 

4. Cataract surgery 

5. Cholecystectomy 

In addition, three diagnoses and one surgical pro- 
cedure have been chosen for which no PSRO im- 
pact is expected. It is predicted that any change which 
has occurred among these would be the same in ac- 
tive and inactive areas. These are: 

Diagnosis or Procedure in which no PSRO Impact is 
Expected 

1. Diabetes 

2. Cerebral hemorrhage 

3. Fracture of the neck of the femur 

4. Prostatectomy 

2.3.3 IMethods 

2.3.3.1 ANALYSIS 

Analyses are performed upon the four surgical pro- 
cedures and one diagnosis to see whether the PSRO 
program has been successful in shifting physicians' 
practice towards less expensive alternatives to the 
prevailing mode of care. For the diagnosis, acute 
myocardial infarction, this reduction would come via 
a shorter ALOS. For three of the four surgical pro- 
cedures, hysterectomy, cataract surgery, and chole- 
cystectomy, a reduction in discharge rate is the de- 
sired outcome. Finally, for the remaining procedure, 
breast cancer surgery, a beneficial change in practice 
would be the reduction in the percentage of radical 
mastectomies compared to all mastectomies (encom- 
passing radical, partial, and simple). 



'M. A. Baum, P. McMenamin, and M. Rudov, Program 
Evaluation Plan: Professional Standards Review 
Organizations, DHEW, CASH, Sept. 1975, pp. 120-122. 



42 



The initial analysis consisted of the production of 
descriptive statistics concerning the chosen diagnoses 
and surgical procedures. For both active and non- 
active PSRO areas, the following utilization statistics 
were produced: 

1. days of care per 1000 aged Medicare 
beneficiaries 

2. discharge rate per 1000 aged Medicare 
beneficiaries 

3. average length of stay. 

These data allow one to look at the trends in utiliza- 
tion by comparing the 1973 and 1977 figures, as well 
as examining the PSRO effect via comparison of the 
active and inactive areas. 

The presentation of descriptive statistics shows 
national and regional diagnostic and procedure spe- 
cific data, with respect to the 1973 and 1977 values for 
the study's utilization measures. These data indicate 
the difference between active and non-active PSROs 
for the use measure under consideration. The data 
appear to be consistent with the literature review 
which suggests that the utilization measures discussed 
here may vary geographically. 

The study's analytical techniques were based on 
the methodology employed in the 1978 Medicare 100 
percent file study, a forced-order multiple regression. 
The general hypothesis is that there has been a 
greater decrease (or lesser increase) in utilization in 
active PSROs compared to inactive PSROs from 1973 
to 1977 for a given diagnosis or procedure. 

As discussed in Section 2.2.3.3, a forced-order 
multiple regression analysis involves, for a given out- 
come variable, entering sets of independent variables 
(base, adjustment, program, and interaction) in an 
equation in order to explain variations in the outcome 
variable. Tests of significance are made at each stage 
of analysis so that the significance of the incremental 
contribution of each set of variables can be assessed. 
This particular regression form is that used in the 
1978 100 percent Medicare study '"> and the 1979 Medi- 
care Impact Analysis. The differences are that 1973 is 
used as a base rate year because the 1 974 20 percent 
discharge file data is incomplete. Also Kansas " is in- 
cluded in this study, although excluded in the 100 
percent file study because of inaccurate coding of 
dead discharges. The definition of active PSRO is 
that used in the 1978 analysis, i.e., a PSRO is con- ' 
sidered active if it had one hospital under binding re- 
view by July, 1977. 

impact estimates are calculated for specific 
diagnoses and procedures where impact is expected, 
for those where no impact is expected, and for all 



'"1978 PSRO Evaluation Section 2.2. 
'^ Kansas is added as an inactive PSRO. 



diagnoses and procedures (referred to below as 
aggregated results)." 

Thus, three types of comparisons are made: (1) ac- 
tive and inactive PSROs are compared to generate 
PSRO impact results; (2) specific diagnostic and pro- 
cedural impact results are compared to aggregate re- 
sults and finally; (3) impact results from diagnoses and 
procedures where impact was expected (impact group) 
are compared to results from diagnoses and proce- 
dures where no PSRO impact was expected (non-im- 
pact group). 

The results of these multiple regression analyses are 
summarized in Section 2.3.4. Those PSRO effects which 
are statistically significant are appropriately indicated 
for each procedure or diagnosis. 

2.3.3.2 DATA SOURCES 

Most of the data come from the four data sources 
described earlier, the Medicare Master Enrollment File, 
the Master Facility Inventory, the Area Resource File, 
and the date on which review started for each PSRO. 
The fifth source is the 20 percent Medicare discharge 
file described here: 

The Twenty Percent Medicare Discharge File " con- 
tains the hospital discharge records of a 20 percent 
sample of all Medicare enrollees. The discharge rec- 
ord includes the identification of the patient, the pa- 
tient's county of residence, and the diagnosis and/or 
procedures recorded at the discharge. All diagnostic 
material is coded using the ICDA-8, modified for 
HGFA use, and surgical codes are assigned using the 
American Medical Association's Current Procedural 
Terminology (CPT) manual, also modified for use in 
HCFA. This file produced the number of live dis- 
charges, dead discharges, and days of care for the 
years 1973 and 1977, both for specific diagnoses and 
procedures ■' and overall. 

■''The impact estimate of percent reduction in the utilization 
measure is calculated with the unstandardized regression 
coefficient. The procedure is the same as discussed in the 
Medicare impact Study. Essentially, the PSRO effect on 
ALOS is the product of the regression coefficient times the 
mean value of longevity. This figure, divided by the estimate 
of ALOS without PSRO review gives the percent reduction 
in the use measure due to PSRO review. This procedure is 
followed for all procedure discharge rates as well. 

"■' Before commencing with this analysis, the 1978 100 percent 
file study was replicated, utilizing the 20 percent file at the 
aggregated level. By looking at the overall DOC, discharge 
rate, and ALOS of this aggregated data, results obtained 
previously should be reproduced. Due to above mentioned 
factors and differences inherent in comparing 20 percent 
and 100 percent samples, the explained variance (R^ value) 
of .838 obtained for the 20 percent sample falls short of 
the R= value of .935 obtained last year for the 100 percent 
sample. However, both cases yielded a statistically 
significant PSRO effect, F=6.3 with the 20 percent file, 
compared with F=47 for the 100 percent file. Thus, it 
appears appropriate to use the 20 percent file in this 
analysis. 

"A good discussion of the 20 percent file is presented in 
"The MEDPAR Report, Medicare Provider Analysis and 
Review," April 1978. 



43 



There are a number of limitations in the data gen- 
erated from the sources described above. According 
to the Institute of Medicine.^^ a large number of diag- 
noses and procedures are not consistently coded into 
the Medicare records, due to a variety of reasons such 
as confusion between principal and primary diagnosis^** 
or incomplete medical records. Reliability appears to 
be an especially acute problem with cases involving 
multiple diagnoses. The Institute of Medicine study 
also showed that the degree of reliability differs ac- 
cording to the individual diagnosis or procedure. 

However, coding is not likely to present a major 
problem in this evaluation for two reasons. First and 
most important, even if there is some inaccuracy in 
coding, there is no reason to suspect that there is 
any association between PSRO activity and the ex- 
tent of inaccuracy." Therefore, the inaccuracy should 
not bias the estimated PSRO effect. Second, most of 
the conditions under consideration are rather straight- 
forward in terms of coding. This would apply to the 
surgeries selected; both elective and for breast cancer. 
However, in the case of the diagnosis selected, the 
lOM study did find that acute myocardial infarction is 
miscoded with a frequency of some 25 percent. How- 
ever, such inaccuracy would not be expected to be 
associated with PSRO activity.'" 

2.3.4 Diagnostic- and Procedure-Specific Results 

2.3.4.1 ACUTE MYOCARDIAL INFARCTION (AMI) 

The utilization measure which the PSRO is hypoth- 
esized to affect for acute myocardial infarction is 
ALOS. Table 25 and Figure 3 indicate that both active 



'"institute of Medicine, op. cit. 

"Primary diagnosis, as defined by Medicare, indicates the 
condition best explaining the hospital stay. The principal 
diagnosis is the Medical Condition which after study is 
determined as responsible for the hospital admission. 

'" It has been suggested that the PSRO impact might enhance 
coding accuracy on claims forms and result in better 
coding in hospitals in active PSROs. However, no evidence 
exists for this argument, either pro or con; as a result this 
is an area in which further research is needed. 

'* Institute of Medicine, op. cit. 



and inactive PSRO areas showed a decrease in ALOS 
from 1973 to 1977, with the active PSROs having a 
longer ALOS in both years. Upon closer examination, 
it can be seen that the difference between the two 
decreased from .72 days (active: 14.91 days vs. inac- 
tive: 14.19 days) in 1973 to .18 days (active: 13.18 days 
vs. inactive: 13.00 days) in 1977. 

The national trends are replicated at the regional 
level (Figure 4). In the Northeast region, which had 
the highest overall ALOS, there was a decrease from 
16.47 to 14.86 days in active PSROs (9.8%) whereas 
the corresponding inactive PSROs show a negligible 
decrease (.1%; 15.20 days in 1973 and 15.18 in 1977). 
Consequently, whereas ALOS for AMI in 1973 was 
1 .27 days greater in active PSRO areas, by 1 977 the 
active areas had an ALOS .32 days lower than the 
inactive areas. This trend of a higher ALOS for active 
PSROs in 1973 and a lower ALOS in 1977 also held 
true in the North Central and West regions, as shown 
in Figure 4. A different trend emerged in the South 
which showed a much smaller differential than the 
other regions. In this region the decrease from 15.14 
to 13.12 days among active PSROs (14.3%) was near- 
ly matched by the decrease in inactive PSROs from 
14.16 to 12.40 days (12.4%). Thus, active PSROs there 
had longer ALOS for both years, .98 days in 1973 and 
.72 days in 1977. 

When forced order multiple regression is applied to 
the data, results substantiate the trends indicated by 
Table 26 that the PSRO has an impact on reducing 
the ALOS for acute myocardial infarction (F^5.8). For 
the aggregated analysis of the 20 percent file, the 
PSRO variable, longevity, accounts for slightly more 
than .5 percent reduction in the ALOS. Looking spe- 
cifically at the results of the acute myocardial infarc- 
tion regression, this figure increases to nearly 2.3 
percent. Thus, the impact of the active PSRO is 
greater on this diagnosis than it is for all diagnoses 
taken together. 

As expected, a multiple regression analysis per- 
formed on the discharge rate did not yield significant 
results. The same held true for the total days of care 
rate, although it is felt that the direction of the re- 
gression coefficient indicates movement in the proper 
direction. 



TABLE 25 

Acute Myocardial Infarction Average Length of Hospital Stay for Aged Medicare 
Beneficiaries for 1973 and 1977, by Active and Inactive PSROs 







Active PSROs 








Inactive PSROs 








1973 


1977 


Change 


Percent 
Change 


1973 


1977 


Change 


Percent 
Change 


Total 


14.91 


13.18 


—1.73 


—11.6 


14.19 


13.00 


—1.19 


- 8.4 


NE 


16.47 


14.86 


—1.61 


— 9.8 


15.20 


15.18 


— .02 


— 0.1 


NC 


15.39 


13.31 


—2.08 


—13.5 


14.51 


13.40 


—1.11 


- 7.6 


South 


15.41 


13.21 


—2.02 


—14.3 


14.16 


12.40 


—1.76 


—12.4 


West 


12.24 


10.89 


—1.35 


—11.0 


12.07 


11.19 


— .88 


— 7.3 



44 



Figure 3 

Average Length of Hospital Stay for Aged Medicare Beneficiaries for 1973 and 1977, 

By Active and Inactive PSROs 

Acute Myocardial Infarction 

Total United States 




YEAR: 


19 73 


,GEND: 


Inactive 




Active 



19 77 



45 



Figure 4 

Average Length of Hospital Stay for Aged Medicare Beneficiaries for 1973 and 1977, 

By Active and Inactive PSROs 

Acute Myocardial Infarction 

By Region 




J u 



_J I L 




46 



TABLE 26 

Discharges per 1,000 Aged Medicare Female Beneficiaries, 
for 1973 and 1977, for Active and Inactive PSROs 



1973 



Active PSROs 

1977 Change 



Percent 
Change 



inactive PSROs 
1973 1977 Change 



Percent 
Change 



Total 


1.15 


.89 


—.26 


—22.6 


1.25 


.92 


—.33 


—26.4 


NE 


1.01 


.80 


—.21 


—20.8 


.98 


.65 


—.33 


—33.7 


NC 


1.39 


.99 


—.40 


—28.8 


1.29 


.98 


—.31 


—24.0 


South 


.95 


.73 


—.22 


—23.2 


1.27 


.96 


—.31 


—24.4 


West 


1.30 


1.05 


—.25 


—19.2 


1.42 


.94 


—.48 


—33.8 



2.3.4.2 HYSTERECTOMY 

It is hypothesized that PSROs would lower hysterec- 
tomy discharge rates. (The use measure in this case is 
discharges per 1,000 aged female Medicare bene- 
ficiaries.) Figure 5, a comparison of active and inactive 
PSROs in 1973 and 1977, shows an overall decrease in 
hysterectomy rates for both active and inactive PSROs, 
with inactive PSROs appearing to show a larger de- 
crease over this time period. Table 26 bears this out 
with the inactive PSROs showing a .33 per 1 ,000 fe- 
males reduction while the active PSROs showed only a 
.26 reduction. Although these reductions seem very 
small, they represent 26.4 and 22.6 percent, respective- 
ly, due to the low discharge rates of the procedure. 

Regionally (Figure 6), all active and inactive PSROs 
show a reduction in discharge rate for hysterectomy 
from 1973 to 1977. In the Northeast, active PSROs 
showed a reduction in discharge rate from 1.01 fe- 
males per 1,000 in 1973 to .80 per 1,000 in 1977 
(20.8%). This compares with inactive PSROs, which 
showed a 33.7 percent decrease in the same time 
period. Correspondingly, the difference between the 
lower rate (inactive PSROs) and higher rate (active 
PSROs) grew from .03 females per 1,000 in 1973 to .15 
females per 1,000 in 1977. In the North Central region, 
active PSROs once again had higher discharge rates 
than inactive PSROs, although the .10 female per 1,000 
in rates in 1973 dropped to a corresponding figure of 
.01 female per 1,000 in 1977. On the other hand, the 
South showed consistently higher rates for inactive 
PSROs. Even though the percentage reduction of 
hysterectomies is similar in active (23.2%) and in- 
active (24.4%) PSROs. The difference between active 
and inactive PSROs decreased from .32 in 1973 to .23 
in 1977, the absolute magnitude of this difference be- 
ing largely responsible for the higher overall rates in 
inactive areas. The West shows a greater change 
among inactive PSROs, where the 1973 figure of 1.42 
females per 1,000 falls to .94 in 1977, a 33.8 percent 
reduction. The active PSROs show a reduction from 
1.30 females per 1,000 to 1.05 per 1,000, a 19.2 percent 
decrease. Thus, although active PSROs had a rate 
which was .12 lower in 1973, inactive PSROs showed a 
.1 1 lower rate for 1 977. 



The results of the multiple regression indicate that 
the PSRO variable has no significant effect on the dis- 
charge rate (F=.004). This is further borne out by the 
fact that the PSRO variable for discharge rate in 
hysterectomies showed an effect in the direction op- 
posite that of the overall PSRO effect on discharge 
rate, accounting for only .2 percent, compared with 
the overall PSRO effect on discharge rate of .8. As 
might be expected, multiple regression analysis on 
DOC rate and ALOS also proved insignificant. Thus, 
it appears that although there has been a decrease 
in discharge rates for hysterectomies, there is no 
evidence that PSRO review was influential in the over- 
all reduction. 

2.3.4.3 CATARACT SURGERY 

Cataract surgery as defined here is the total of two 
coded procedures, lens extraction and the Von Graffe 
procedure, whose code includes lens implantation. As 
is the case with all the surgical procedures (except 
breast cancer surgery) of concern, the utilization 
measure for expected PSRO impact is the discharge 
rate. Figure 7 and Table 27 indicate both active and 
inactive PSROs had higher discharge rates for cataract 
surgery in 1977 than in 1973. This fact is borne out In 
Table 27 where the difference between active and 
inactive PSROs in 1977 is nearly seven times that 
which occurred in 1973. Once again, there is a wide 
variety of trends which occur when looking at cataract 
discharge rates across regions (Figure 8). In the 
Northeast, active PSROs showed consistently higher 
discharge rates than inactive areas. The rates for 
active PSROs increase from 9.40 per 1,000 to 10.09 
(7.3%) while the inactive PSROs showed a negligible 
decrease from 8.37 to 8.33 (.5%). The absolute differ- 
ence between active and inactive PSROs increased 
from 1.03 per 1,000 in 1973 to 1.66 in 1977. In the 
North Central region, the active PSROs again had 
higher discharge rates than the corresponding inactive 
PSROs. However, the discharge rate for inactive 
PSROs increased at a more rapid pace, so the differ- 
ence between actives and inactives had decreased 
from 0.52 in 1973 to 0.19 in 1977. The South showed 
the largest absolute difference between active and 



47 



Figure 5 

Hysterectomy Discharges per 1,000 Aged Medicare Female Beneficiaries for 

1973 and 1977, for Active and Inactive PSROs 

Total United States 



o 
o 
o 



w 

60 

o 
m 

•H 
Q 



1.35 



1.30 



1.25 



1.20 - 



1.15 - 



1.10 - 



1.05 - 




1.00 - 



YEAR: 1973 

LEGEND: Inactive 

Active 



1977 



48 



Figure 6 

Hysterectomy Discharges per 1,000 Aged Medicare Female Beneficiaries for 
1973 and 1977, for Active and Inactive PSROs 




I L 



J 1 1 L 




/ - 



-I I I L. 



ooo'T ma saDHVHDsia 



000 'T uaa saowrosia 




J i_ 



_l L. 




I 1 1 L. 



J 1 L 



0^ CO 



^£) ir\ 



COO T aaj saoavHDSia 



000 T aaa saowHosia 



49 



Figure 7 

Cataract Surgery Discharges per 1,000 Aged Medicare Beneficiaries for 

1973 and 1977, By Active and Inactive PSROs 

Total United States 



11. Of 



10.5 - 



10.0 - 



o 
o 
o 






CO 




8.0 



YEAR 


19 73 


.EGEND: 


Inactive 




Active 



19 77 



50 



Figure 8 

Cataract Surgery Discharges per 1,000 Aged Medicare Beneficiaries for 
1973 and 1977, By Active and inactive PSROs 



000 'T Had saoavHDSia 



000 'T Md saoavHDsia 




000' T aad saoOTHDSia 



000 'T Had SaOHVHDSIC 



51 



TABLE 27 

Cataract Surgery Discharges per 1,000 Aged Medicare Beneficiaries, 
for 1973 and 1977, by Active and Inactive PSROs 



1973 



Active PSROs 

1977 Change 



Percent 
Change 



Inactive PSROs 
1973 1977 Change 



Percent 
Change 



Total 


9.03 


9.54 


.51 


5.6 


9.17 


10.51 


1.34 


14.6 


NE 


9.40 


10.09 


.69 


7.3 


8.37 


8.33 


-.04 


—0.5 


NC 


8.54 


9.62 


1.08 


12.6 


8.02 


9.43 


1.41 


17.6 


South 


7.86 


7.96 


.10 


1.3 


10.28 


12.26 


1.98 


19.3 


West 


9.88 


9.97 


.09 


0.9 


9.62 


10.24 


.62 


6.4 



inactive PSROs. While active PSROs showed consist- 
ently low discharge rates for cataract surgery (7.86 in 
1973 to 7.96 in 1977, an increase of 1.3%) the inactive 
PSROs had an initially high rate of 10.28 discharges 
per 1,000 population in 1973 which increased to 12.26 
in 1977 (19.3%). Thus, while the difference between 
active and inactive PSROs was 2.42 per 1,000 in 1973, 
it nearly doubled to 4.30 in 1977, largely affecting the 
national figure. In the West, another trend emerged 
where inactive PSROs had a lower rate by .26 per 
1,000 in 1973, but active PSROs had a rate which was 
.26 per 1,000 lower in 1977. 

The results of the multiple regression performed on 
cataract surgery data indicate a significant impact on 
the discharge rate for active PSROs, compared to in- 
active PSROs (F=6.0). The implication of the regres- 
sion results is that PSRO review yields a 7.9 percent 
decrease in the discharge rate. This value compares 
favorably to the aggregated discharge rate value of 
.8 percent and indicates that the PSRO impact is in- 
deed strong for this procedure. Thus, even though the 
ALOS impact proves insignificant in multiple regres- 
sion analysis, the strength of the discharge rate anal- 
ysis yields an F value of 4.2 for the DOC rate, which 
also is significant. Thus, the 4.6 percent impact of the 
PSRO variable on TDOC for cataracts is greater than 
the overall impact of 3.3 percent. 

2.3.4.4 CHOLECYSTECTOMY 

The hypothesized point of reduction due to the 
PSRO impact for cholecystectomy, or surgical excision 



of the gall bladder, was the discharge rate. Figure 9 
and Table 28 indicate that while the discharge rate 
rose from 3.19 discharges per 1,000 in 1973 to 3.46 
discharges per 1,000 in 1977 among inactive PSROs, 
the rate fell from 3.07 discharges per 1,000 to 2.95 for 
their active counterparts over this same time period. 
The difference between active and inactive PSROs in- 
creased from .12 per 1,000 in 1973 to .51 per 1,000 
in 1977. 

Figure 10 indicates that a variety of trends emerge 
when looking at the four regions individually. In the 
Northeast region, active PSROs show a decrease in 
rate per 1,000 of 6.7 percent from 1973 to 1977 while 
the inactive PSROs increased 2.9 percent in the same 
time frame. The initially higher figure for active PSROs 
in the Northeast of .34 discharges per 1 ,000 decreased 
to .05 discharges per 1,000 in 1977. Active and inac- 
tive PSROs followed a similar trend in the North 
Central region, with active PSROs showing an increase 
from 3.30 to 3.47 discharges per 1,973 (5.2%), while 
inactive PSROs showed an increase from 3.32 to 3.43 
discharges per 1,000 (3.3%). Obviously, the differences 
by year were slight, with active PSROs having a rate 
.02 discharges per 1,000 lower in 1973 and .04 higher 
in 1977. The South showed consistently lower rates for 
active PSROs, with the difference between actives and 
inactives widening from .41 per 1,000 in 1973 to 1.21 
per 1,000 in 1977. Active PSROs also attained lower 
discharge rates in the West region. However, the dif- 
ference between the two rates >7ecreased from 52 per 
1,000 discharges to .38 per 1,0U0 discharges over the 
time period. 



TABLE 28 

Cholecystectomy Discharges per 1,000 Aged l\1edicare Beneficiaries, 
for 1973 and 1977 — Active and Inactive PSROs 







Active PSROs 








Inactive PSROs 








1973 


1977 


Change 


Percent 
Change 


1973 


1977 


Change 


Percent 
Change 


Total 


307 


2.95 


—.12 


—3.y 


3.19 


3.46 


.27 


8.5 


NE 


3.12 


2.91 


—.21 


—6.7 


2.78 


2.86 


.08 


2.9 


NC 


3.30 


3.47 


.17 


5.2 


3.32 


3.43 


.11 


3.3 


South 


2.69 


2.55 


—.14 


—5.? 


3.10 


3.76 


.66 


21.3 


West 


3.09 


2.87 


—.22 


—7.1 


3.61 


3.25 


—.36 


—10.0 



52 



Figure 9 

Cholecystectomy Discharges per 1,000 Aged Medicare Beneficiaries for 
1973 and 1977, By Active and Inactive PSROs 
Total United States 



4.0 



3.5 



3.0 



o 
o 
o 



(4 
Pi 






2-.0 



1.5 



1.0^ 




i 



YEAR 
LEGEND 



53 



Figure 10 

Cholecystectomy Discharges per 1,000 Aged Medicare Beneficiaries for 
1973 and 1977, By Active and Inactive PSROs 





' f 


_ 




' 






/ 






/ 






' 






' 






/ 




H 


/ 










W 


/ 




3 


/ 




L... 


/ / 

1 1 





000 'T Had saoHVHosia 



000' T aaa saonvHOsia 



000 'T Had saoHVHosia 



000' T Had saoHVHOSia 



54 



The results of the multiple regression performed 
upon cholecystectomy data indicate a significant re- 
duction in discharge rate due to the PSRO effect 
(F=7.1). This, in turn, accounts for a 7.8 percent re- 
duction of the discharge rate for cholecystectomy due 
to the PSRO effect, compared with the much smaller 
figure of .8 percent for the aggregated discharge rate. 
This provides evidence for the impact of the PSRO 
effect when looking at this particular situation. 

Although the results of the multiple regression for 
ALOS proved insignificant in regard to cholecystec- 
tomy, the TDOC rate was found to be highly significant 
(F^8.5). The reduction in days of care due to PSRO 
activity is estimated to be 10.2 percent, substantially 
higher than the 3.3 percent figure found for aggregated 
data. 

2.3.4.5 BREAST CANCER 

Within the 20 percent Medicare file, there are three 
types of mastectomies coded: radical, partial, and 
simple. In the literature review pertaining to breast 
cancer (Section 2.4.2.7), it was suggested that PSRO 
review might have the effect of shifting breast sur- 
geries from radical to simple or partical mastectomies. 
Therefore, the ratio of radical to combined radical, 
partial, and simple mastectomies was explored. As is 
the case with hysterectomies the population at risk 
in this analysis consists of the aged female population. 

The overall percentage of mastectomies which are 
radical is consistently higher in active PSROs than 
inactive PSROs (Figure 11). Active PSROs show a 
decrease from 37.9 percent in 1973 to 36.0 percent 
in 1977 (5.0% relative change) while inactives show 
a decrease from 37.2 percent to 35.2 percent, a similar 
change of 5.4 percent. Regionally, Figure 12 and Table 
29 indicate a variety of trends occurring in the indi- 
vidual regions. The Northeast shows the percentage 
of radical mastectomies to total mastectomies rising 
for both active and inactive PSROs. The actives in- 
creased from 36.8 percent to 38.0 percent (a relative 
change of 3.3%) whereas inactive PSROs had an in- 
crease from 34.8 percent to 36.7 percent (5.5% rela- 



tive change). Thus, the gap between the two narrowed 
as inactives increased at a larger rate. The trend was 
completely different in the North Central region, where 
the percentage among actives decreased from 39.2 
to 33.7 percent (relative change of 14.%) while in- 
active PSROs showed an increase from 34.5 percent 
to 35.7 percent (a relative change of 3.5%). Therefore, 
one can see in Figure 12 that while active PSROs had 
a higher rate of radical mastectomies in 1973, they 
had a lower rate compared to inactive PSROs in 1977. 
The South experiences still another pattern, where 
both active and inactive PSROs showed a decreasing 
rate of radical mastectomies. Active PSROs decreased 
from 34.1 percent to 32.7 percent, a relative change 
of 4.1 percent whereas inactive PSROs showed a 
larger decrease, from 37.4 percent to 32.9 percent, a 
relative change of 12.1 percent. Thus, although active 
and inactive PSROs differed by over 3 percent in rate 
of radical mastectomies compared to all mastectomies 
in 1973, this figure dwindled to less than 1 percent 
in 1977. The West showed a trend similar to the South, 
with both active and inactive PSROs showing a de- 
crease, and the difference between actives and inac- 
tives decreasing from 1973 to 1977. 

The multiple regression comparing active to inac- 
tive PSROs yielded a significant change in the rate of 
radical mastectomies compared to total mastectomies 
(F=6.1). This seems surprising when one considers 
the relationship illustrated in Figure 11. To examine 
this situation in greater detail, the active PSROs were 
looked at in terms of their age. 

Table 30 and Figure 13 both illustrate that the rate 
of radical mastectomies varies greatly with PSRO age. 
Thus, it appears that although the overall active effect 
appears similar to that found in inactive PSROs, older 
PSROs (25+ months) appears to be more successful 
in showing a rate reduction, whereas the opposite 
effect appears in the newer organizations. It should be 
noted that these figures indicate the status of the 
PSROs at specific points in time, and does not imply 
that PSROs go from rate increase to rate decrease 
with regard to radical mastectomy as they grow older. 



TABLE 29 

Ratio of Radical Mastectomies to all Mastectomies per 1,000 

Aged Female Medicare Beneficiaries, for 1973 and 1977 — 

Active and Inactive PSROs 



1973 



Active PSROs 

1977 Change 



Percent 
Change 



Inactive PSROs 
1973 1977 Change 



Percent 
Change 



Total 


37.9 


36.0 


—1.9 


—5.0 


37.2 


35.2 


—2.0 


—5.4 


NE 


36.8 


38.0 


1.2 


3.3 


34.8 


36.7 


1.9 


5.5 


NC 


39.2 


33.7 


—5.5 


—14.0 


34.5 


35.7 


—1.2 


3.5 


South 


34.1 


32.7 


—1.3 


—4.1 


37.4 


32.9 


—4.5 


—12.0 


West 


41.4 


37.9 


—3.5 


—8.5 


46.9 


40.0 


—6.9 


—14.7 



55 



Figure 11 

Ratio of Radical Mastectomies to all Mastectomies per 1,000 Aged Female Medicare 
Beneficiaries for 1973 and 1977, By Active and Inactive PSROs 
Total United States 



o 
o 
o 



Pi 
w 

o 

M 
H 




.30 



YEAR 


19 73 


LEGEND: 


Inactive 




Active 



19 77 



56 



Figure 12 

Ratio of Radical Mastectomies to all Mastectomies per 1,000 Aged Female 
Beneficiaries for 1973 and 1977, By Active and Inactive PSROs 




I I I I I I L 

cj\ oo r-^ \0 in ^ fo 




000 'T sad oiivH 



000 T S3d oiiva 





\ \ 


. 




\ \ 






\ \ 






\ \ 






\ \ 






V \ 




H 


\ \ 




m 






< 


\ \ 




g 


\ \ 




OS 




o 






z 


\ \ 
\ \ 
\ \ 
\ \ 
\ \ 
\ \ 
\ \ 
\ \ 




1 1 


1 1 1 1 1 


1 




000 T aad oiiva 



000 T Had oiivu 



57 



Figure 13 



,42 



.41 - 



.40 



39 - 



o 
o 
o 



w 
o 

M 
H 



.38 



37 



36 ^ 



.35 



34 



Ratio of Radical Mastectomies to all Mastectomies per 1,000 Female Beneficiaries 
for 1973 and 1977, By Active and Inactive PSROs By PSRO Age 
Total United States 



/ 



PSRO, 1-12 




PSRO, 13-24 



Inactive PSRO area 



PSRO, 25+ nonths old 



Tionths old 



months old 



X 



YEAR 19 73 



1977 



58 



TABLE 30 

Ratio of Radical Mastectomies to ail Mastectomies per 

1,000 Aged Female Medicare Beneficiaries in 1973 

and 1977 by Active PSROs, Total United States 



Inactive 



1-12 
Months 



Active 

13-24 

Months 



25+ 
Months 



1973 
1977 



37.2 
36.9 



36.5 
41.6 



38.5 
38.1 



40.9 
35.3 



2.3.4.6 NON-IMPACT GROUP COMPARED TO 
IMPACT GROUP 

For the one surgical procedure (prostatectomy) and 
the three diagnoses (diabetes, cerebral hemorrhage, 
and fracture of the neck of the femur), the utilization 
measure of interest was the discharge rate. Table 31 
illustrates that multiple regression analysis found, in 
every case, there was no significant PSRO effect in 
this non-impact group. 

Therefore, as expected, active PSROs were not 
significantly different from inactive PSROs regarding 
the impact on discharge rates of this group of 
conditions. 

2.3.4.7 CONCLUSIONS 

An analysis has been performed upon four surgical 
procedures (hysterectomy, cataract surgery, cholecy- 
stectomy and breast cancer surgery) and one diag- 
nosis (acute myocardial infarction) to see whether 
active PSROs have had an impact upon utilization, 
relative to inactive PSROs. In every case except one, 
hysterectomy, the results of the forced order multiple 
regression indicated statistically significant results 
which supported the hypotheses. 

Looking at the procedures and diagnosis, two of 
the cases (acute myocardial infarction and hysterec- 



tomy), both had active and inactive PSROs showing 
declining rates for the time period 1973 to 1977. When 
looking at acute myocardial infarction, it was found 
that the utilization measure of hypothesized impact, 
ALOS, decreased at a significantly faster rate for ac- 
tive PSROs. Thus, in this case, the active PSRO ap- 
pears to facilitate the acceptance of the shorter ALOS 
described in the literature review. For hysterectomies, 
there seemed to be no significant difference in the 
rate of decrease between active and inactive PSROs 
regarding discharge rate. There had been a general in- 
crease in cataract surgery in this time period, due 
largely to an increasing number of operations per 
surgeon.'" As a result, both active and inactive PSROs 
showed increases in discharge rates. However, the in- 
crease was smaller for active PSROs. Perhaps the 
most graphic example of the PSRO effect is cholecy- 
stectomy, where active PSROs show a decrease in 
discharge rates from 1973 to 1977, while inactive 
showed an increase in discharge rate in the same time 
period. Both inactive and active PSROs showed a re- 
duction in percentage of mastectomies which were 
radical. However, the variation among active PSROs by 
duration of binding service was significant. In this 
case, the length of PSRO activity is important be- 
cause while PSROs of intermediate age appeared to 
parallel the trends of the inactive PSROs, active 
PSROs less than a year old show an increase in rate 
of radical mastectomies compared to all mastectomies 
from 1973 to 1977, whereas those two years or older 
showed a decrease. 

In certain instances, the differences between active 
and inactive PSROs decreased from 1973 to 1977, 
prompting the possibility of a "regression to the 
mean" effect. However, in both the national and re- 
gional presentations of the data, it can be seen that 
this is only one phenomenon. In many cases, a diver- 
gence or cross-over effect (where for instance an ac- 
tive PSRO has higher utilization in 1973 but lower in 



" J. Taylor, op. cit. 



TABLE 31 

Comparison of Statistical Significance Between 
Hypothesized Impact and Non-Impact Groups 





Impact Group 






Non-Impact Group 


Name 


Utilization 
measure 


F value 


Name 


Utilization 
measure 


F value 


Acute myocardial 

infarction 
Hysterectomy 
Cataract surgery 

Cholecystectomy 
Radical mastectomy 


Length of stay 

Discharge rate 
Discharge rate 

Discharge rate 
Percent of all 
mastectomies 


5.8 

.004 (ns) 
6.0 

7.1 
6.1 


Diabetes 

Cerebral hemorrhage 
Fracture of the neck 

of the femur 
Prostatectomy 


Discharge rate 

Discharge rate 
Discharge rate 

Discharge rate 


.68 (ns) 

2.7 (ns) 
2.3 (ns) 

1.7 (ns) 



59 



1977) takes place. Therefore, there is no consistent 
convergence between active and inactive PSROs to 
substantiate the hypothesis of "regression to the 
mean." 

A decrease in utilization effected by the PSRO pro- 
gram was found in the 1978 100 percent analysis. An 
important question is whether the decrease is the re- ^ 
suit of an impact in certain procedures and diagnoses, 
or whether it has been more general. The results of 
the diagnosis and procedures analyzed here, for which 
active PSRO effect was hypothesized, indicated PSRO 
impact may be attributed to specific diagnoses and 
procedures /iather than being a general effect. As a 
means of comparison, the three diagnoses (cerebral 
hemorrhage, fracture of the neck of the femur, and 
diabetes), and one procedure (prostatectomy) for 
which no impact was expected showed no statisti- 
cally significant effect with respect to PSRO activity. 

On the basis of these ten procedures and diagnoses, 
it does appear that the PSRO program has been rela- 
tively successful in having an impact in specific areas 
where impact was expected to occur, while having no 
impact where none was expected. However, these re- 
sults should be interpreted with caution. First, the 
provider-based data which are utilized here do not 
account for patient migration. Second, because the 
data are from a sample, there may be sampling error. 
Third, the diagnoses selected were generated by the 
evaluation staff and were not necessarily topics of 
interest to the PSROs. Thus, although several of the 
selected conditions were among the list approved in 
1975 by the National Professional Standards Review 
Council for evaluation of the PSRO program in terms 
of quality of care, utilization, and cost of care,^' priori- 
ties assigned by individual PSROs are unknown. 
Fourth, the Institute of Medicine study indicates sus- 
pected inaccuracies in the coding of the data; al- 
though there is no reason for believing that coding 
error would differ across the PSROs, errors in cod- 
ing would obviously weaken the confidence that can 
be placed in the statistical results. 

Thus, given the limited nature of the data and the 
potential problems of bias in outcome measurement, 
the results of these analyses suggest that the PSRO 
impact is significant in medically identifiable problem 
areas where it is thought to occur. 

2.4 Medicaid IHospitai Utilization Rate 
Study (IVIedicaid Rate Study) 

2.4.1 Objectives 

The 1977 and 1978 PSRO evaluations focused pri- 
marily on the impact of PSRO review on Medicare 
beneficiary hospital utilization at the national and re- 
gional level. No such analyses were conducted for 
Medicaid beneficiaries because no data exist at the 
national level to support Medicaid analyses within the 
context of the PSRO evaluation. 



' M. A. Baum et al., op. cit. 



This study explores the potential of using existing 
data sources to determine the impact of the PSRO 
program on Medicaid beneficiary hospital utilization. 
The 1979 PSRO evaluation and its predecessors stress 
the importance of using standardized measures of 
utilization such as days of care and discharge rates 
per 1,000 eligibles and average length of stay in the 
evaluation of the PSRO program impact on Federal 
beneficiaries. Utilization rate measures are comprised 
of utilization (numerator) and population-at-risk eligi- 
bility (denominator) data. Successful Medicaid rate 
analyses at the national level have not yet been con- 
ducted because numerous problems exist with both 
Medicaid numerator and denominator data. This study 
uses PSRO Hospital Discharge Data Set (PHDDS) 
utilization data as numerator data and Medicaid State 
agency eligibility data as denominator data in the 
construction of Medicaid use rates. 

The primary purpose of this study is to develop 
methods for analyzing and comparing Medicaid rate 
changes. Four PSRO case studies are presented to 
explore the strengths and weaknesses of proposed 
methodologies and to investigate basic hypotheses. 
The findings include a limited discussion of rate trends 
within the four case study PSROs. 

The study's first primary objectives are shown in 
Table 32: 



TABLE 32 
Medicaid Rate Study Objectives 

1. Explore the use of PHDDS utilization data as 
numerators for Medicaid rate analyses; 

2. Explore the use of state Medicaid agency 
eligibility data as denominators for Medicaid 
rate analyses; 

3. Develop appropriate methods for the linking of 
PHDDS data to Medicaid agency eligibility data 
in rate construction; 

4. Calculate Medicaid utilization rates and analyze 
trends for each of the PSRO case study areas; 

5. Recommend further analyses, specifying data 
needs and accompanying analytical framework. 



This study does not attempt to compare Medicaid 
rates among PSROs nor does it provide evidence for 
causal relationships between PSRO activities and 
changes in hospital utilization because adequate con- 
trols were not available. That is, the PHDDS includes 
no data for pre-PSRO activity or for non-PSRO areas. 
Any observed change in hospital utilization within a 
PSRO can only be said to occur simultaneously with 
PSRO review. However, a number of summary state- 
ments are made regarding observed trends in the 
individual case studies. 

The study begins with a presentation of objectives 
and a brief discussion of the data sources. Tliese sec- 
tions are followed by a discussion of the key issues 
and problems in developing Medicaid rates. The dis- 



60 



cussion continues with sections on methodology, case 
study findings and general results. The study con- 
cludes with detailed recommendations for the future 
Medicaid rate development activities that will be re- 
quired to support analyses of PSRO review impact on 
Medicaid beneficiaries. 

2.4.2 Data and Case Study Area Selection 

As mentioned above, utilization rate data are de- 
veloped using discharge abstract data from the 
PHDDS as numerator quantities and eligibility data 
from the state agencies as denominator quantities. The 
following discussion describes both numerator and 
denominator data elements. 

2.4.2.1. PHDDS UTILIZATION DATA 

The PHDDS was chosen as a source of utilization 
data for this study because it had several advantages 
over other possible sources. First, the PHDDS is the 
only source of hospital discharge data for Medicaid 
beneficiaries available from more than a handful of 
PSRO areas. The uniformity of discharge abstracting 
in the PHDDS generates a larger collection of avail- 
able data elements than any other source. Moreover, 
consistency of diagnostic coding should exceed that 
of any other Medicaid data source. Because the file 
is maintained at the Federal level, complete data 
files were similar in structure and inexpensive to ob- 
tain. Finally, PHDDS data files are readily available 
from those PSROs that have a PHDDS system in 
place. 

Hospital discharge abstract data are collected by 
HCFA for each active PSRO area on a quarterly basis. 
This data set includes information on all Medicaid 
discharges over which the PSRO has review authority. 
Data elements recorded on the file include: 

• Patient age 

• Patient sex 

• Zip code of patient residence 

• Diagnosis codes 

• Procedure codes 

• Expected source of principal payment (Medicare, 
Medicaid, etc.) 

• Admission date 

• Discharge date 

• Discharge status 

2.4.2.2 MEDICAID ELIGIBILITY DATA 

Because there is no central Federal source of data 
on Medicaid eligibles, the state Medicaid agency for 
each of the PSROs participating in this study was 
asked to supply counts of Medicaid eligibles directly 
from their administrative files. Given the short time 
frame for this study, the data request was limited to 
point-in-time estimates of eligibles for the first month 
of each quarter throughout the study period. To facili- 
tate the construction of meaningful rates, counts of 
eligibles were requested to be broken out in the 
following ways: 



1 . Program Category of Eligibility: 

• Old Age Assistance (OAA) 

• Aid to Families with Dependent Children 
(AFDC) 

• Aid to Blind (AB), and 

• Aid to Permanently and Totally Disabled 
(APTD) 

2. Financial Category of Eligibility: 

• Categorically Needy 

• Medically Needy 

3. Demograpfiically: Age by Sex 

4. Geographically: County or Zip Code of 
Residence 

5. Time Period: 1 976 to 1 978 

Considerable variation exists among the states in 
both the availability and level of detail of their eligi- 
bility counts. A detailed discussion of the Medicaid 
eligibility data that were actually acquired for this 
effort is included in the contractor's report.®* 

2.4.2.3 CASE STUDY SELECTION 

The principal consideration in the selection of 
PSRO areas for the study was the availability of com- 
plete (or substantially complete) PHDDS data for 
1976-1978 so that changes in utilization taking place 
over an extended period could be identified and ex- 
amined. A complete file of PHDDS discharge data for 
1976-1978 was reported for only three PSRO areas. 
Data for 1977-1978 were acquired from one additional 
PSRO area. Consequently, analyses were conducted 
on twelve quarters of data in three PSRO areas and 
eight quarters in the fourth. For reasons of confiden- 
tiality, these four case study PSRO areas are not 
identified by name in the study. 

2.4.3 Issues and Problems in Developing Rates 

2.4.3.1 COMPLEXITIES OF THE MEDICAID 
PROGRAM 

The Medicaid program, as established under Title 
XIX of the Social Security Act, is a complex decen- 
tralized program. Medicaid is not a single Federal 
program, but is rather an "umbrella" of state programs 
that provide care for four major medical aid groups: 
Aid to Families with Dependent Children (AFDC), Aid 
to the Blind (AB), Old Age Assistance (OAA), and Aid 
to the Permanently and Totally Disabled (APTD). 
Eligible individuals are further categorized as "medi- 
cally needy" or "categorically needy." These popula- 
tions have varied compositions with respect to health 
status, age, and both the type and amount of use of 
specific medical services. The distribution of indi- 
viduals in these groups varies from state to state and 
over time within state. Eligibility criteria also vary 
from state to state as do benefit packages (both in 



M. O'Brien, F. Jagger, et a!., "Evaluation of Trends of 
Medicare and Medicaid Utilization," Vol. 1, SysteMetrics, 
Inc., January 15, 1980. 



61 



terms of the scope and level of services provided). It 
is difficult to develop consistent national Medicaid 
utilization rates because of these program char- 
acteristics. 

Two previous analyses highlight these issues. The 
Colorado Case Study of the 1977 DHEW PSRO Evalua- 
tion **2 stressed difficulties in obtaining consistent 
eligibility statistics and in removing v\/ide fluctuations 
in use rates caused by individuals with transient Medi- 
caid program eligibility. In a study of utilization for 
the New Mexico Experimental Medical Care Review 
Organization (EMCRO), Brook «3 argues that stable 
Medicaid use rates can only be derived by isolating 
cohorts of individuals who maintain continuous eligi- 
bility over the study period. This is impossible with 
PHDDS data because patient identifiers cannot be 
readily linked to eligibility histories. Brook's conclu- 
sion, however, does suggest the difficulty of the task 
at hand. 

In the next two sections, issues are identified and 
potential techniques to adjust or compensate for 
those problems are presented. Numerator and de- 
nominator issues are discussed separately. A summary 
of these issues is presented in Table 33. 

2.4.3.2 PHDDS DATA (NUMERATOR) ISSUES 

2.4.3.2.1 Patients by Program Category 

As noted above, existing analyses of Medicaid pa- 
tients have shown that the age, sex, and health status 
differences among the various Medicaid program 
categories (AFDC, OAA, AB, and APTD) are associated 
with very different patterns of hospital utilization. 
Likewise, utilization rates of the medically needy re- 
cipients are expected to be higher than the cash grant 
recipients in the same eligibility category. As a result, 
a trend analysis of aggregate utilization rates in the 
Medicaid populations is suspect unless it is possible to 
control for changes in the program category composi- 
tion of the Medicaid population. Because PHDDS data 
do not enable control for changes in program category 
mix, trends presented below are unadjusted for pos- 
sible mix differences. This is a major problem with 
PHDDS Medicaid data analysis. 

2.4.3.2.2 Patient Migration 

In Medicare rate analyses, the denominator popula- 
tion (the number of enrollees residing in a PSRO area) 
must be adjusted for patient migration into or out of 



TABLE 33 
Summary of Key Issues in Medicaid Rate Construction 



*^PSRO: An Evaluation of the Professional Standards Review 
Organization, Volume 5: A Comprehensive Case Study: 
The Colorado PSRO Experience, No. 77-12, OPEL, HSA, 
PHS, DHEW, October 1977, pp. 109-116. 

*< Brook, R. H., Williams, K. N., Rolph, J. E., and Mori, B.: 
Controlling the Use and Cost of Medical Services: The 
New Mexico Experimental Medical Care Review 
Organization — A Four Year Case Study, Medical Care, 
Vol. 16, No. 9, September 1978. 





Resolved 


Issue 


for this 
Analysis? 




Yes 


No 


PHDDS (Numerator): 






1. Program category 




X 


2. Patient Migration 






a. In-migrants 


X 




b. Out-migrants 




X 


3. Inaccurate reporting of discharges 






a. Inclusion of "state only" 


x 




patients 




. 


b. Retroactive eligibility 




X 


c. Indeterminate status 




X 


d. Newborn and obstetrical 


X 




patients 






4. Dual eligibility ("crossovers") 


X 




5. Patients who die in hospital 


X 





Eligibility (Denominator): 

1 . Data variability among states 

a. Composition of eligibility 
categories 

b. Data retention 

c. Specificity of available data 

2. Methods of counting eligibles 

a. Participation of the eligible 
"pool" 

b. Turnover of eligible population 

3. Missing data 



a PSRO area. Deacon ^' has estimated a migration ad- 
justment factor for the Medicare denominator popula- 
tion for each PSRO included in the Medicare Impact 
Study. 

The level of patient migration and changes in mi- 
gration patterns over time most likely also have an 
effect on Medicaid utilization rates. However, Deacon's 
method could not be adapted to PHDDS Medicaid rate 
analyses because such calculations would require 
PHDDS and eligibility data for neighboring PSRO 
areas.85 Generally, these data are not available for 
enough neighboring PSRO areas to make the cal- 
culation feasible. 

The Medicaid rates presented below exclude dis- 
charges caused by "inmigration," but do not adjust 
for "out-migration." The rate formula used is: 



" R. Deacon, J. Lubitz, M. Gornick, and M. Newton, "Analysis 
of Variations in Hospital Use by Medicare Patients in 
PSRO Areas, 1974-1977," Health Care Fmancing Review, 
Vol. 1, Summer 1979. 

^ The Deacon method uses information from all other PSROs. 
Migration adjustments for neighboring PSROs would 
approximate the Deacon approach. 



62 



Medicaid Discharge Rate= 
Total Discharges Less 



In-Migrant Discharges 



Medicaid Eligibles in the PSRO 

Although the level of out-migration for Medicaid 
has not been measured, it has been hypothesized 
that migration across state lines may be less frequent 
for Medicaid patients than for Medicare patients, be- 
cause Medicaid restricts the locations where state 
residents may receive covered services. 

2.4.3.2.3 Inaccurate Reporting of Discharges 

In theory, each PSRO is required to complete 
PHDDS abstracts on all patients discharged who are 
covered under Title XVIII (Medicare), Title XIX (Medi- 
caid) or Title V (Maternal and Child Health). In addi- 
tion, the PSRO may contract to do review of other pa- 
tients, which also may be included on the PHDDS. 
in practice, the PHDDS Medicaid data often contain 
non-Medicaid beneficiary records, while failing to 
include the entire "mandated" population. These 
inclusions and exclusions can limit the usefulness of 
the PHDDS of calculating Medicaid utilization rates. 
The categories of patients for which there are incon- 
sistencies in reporting and coding across PSROs 
include: 

• "State only" patients 

• Patients with retroactive eligibility 

• Patients with indeterminate status 

• Newborn and obstetrical discharges 

State Only Patients 

Many states offer special Medical Assistance pro- 
grams that are administered by the state agencies, but 
are not funded by Federal Medicaid dollars. In many 
PSROs, these "state only" patients are included in 
PHDDS data, as Medicaid discharges, despite the fact 
they should be excluded. Since the Medical Assistance 
card for these patients is often indistinguishable from 
that of Federal Medicaid patients, the PSRO has no 
way to isolate these patients. In some states, the Medi- 
caid identification number which appears on the Medi- 
cal Assistance card contains a program category 
code which may be used to separate Federal from 
"state only" patients. Often, however, the identifica- 
tion number is the Social Security number, which does 
not separate the two. 

Even when the Medicaid ID number can be used to 
distinguish Federal from state programs, some PSRO 
areas continue to review"state only" patients. In other 
cases, the PSRO originally abstracted "state only" 
cases, but now includes only Federal patients on the 
PHDDS. In this study it was possible to compensate 
for the inclusion of "state only" patients by inflating 
the Medicaid denominator with "state only" eligibles 
in PSROs where review of these patients is taking 
place. 



Retroactive Eligibility 

The PHDDS contains abstracts for all patients whose 
"expected" source of payment is Medicaid. Since 
hospital admission can often be the impetus to ap- 
plying for a Medicaid card, especially among the 
medically indigent or the "spend-down" population, 
patients whose eligibility is confirmed after hospital 
discharge may or may not appear on the PSRO's 
PHDDS. PSRO procedures for inclusion of these pa- 
tients vary, as does the percentage of patients who 
fall into this class. It appears that when PSROs utilize 
hospital discharge abstract services it is more difficult 
to include these retroactive eligibles. Where the PSRO 
has its own data system, utilization review coordinators 
are more easily able to follow up these cases in the 
hospital billing offices, and are able to update the 
PHDDS accordingly. However, for this analysis, there 
was no way to completely compensate for patients 
whose eligibility was confirmed retroactively. 

Indeterminate Status 

Besides excluding patients whose eligibility status 
is indeterminate at hospital discharge but confirmed 
later, a PSRO may err by including on the PHDDS, as 
Medicaid, other patients whose eligibility status is 
indeterminate at the time of discharge and who later 
are shown to be ineligible. The relative proportion of 
these patients is not stable at different points in time 
or across areas. No corrections to rate calculations 
were possible for these patients. 

Newborn and Obstetrical Patients 

Although HGFA has requested the inclusion of all 
newborn discharges in the PHDDS, some PSROs 
routinely reported newborns and other PSROs did not 
report newborns for the entire study period. Still 
other PSROs chose to report data for only the new- 
borns that were reviewed by the PSRO. In addition, 
PSROs were not consistent in their use of diagnostic 
codes for these discharges. Because of these incon- 
sistencies within and among PSRO areas, newborns 
were excluded from the analyses. Similarly, because of 
reporting inconsistencies, obstetrical discharges were 
excluded from the analysis in one area. 

2.4.3.2.4 Patients with Dual Eligibility 

One category of eligibles which are in fact coded 
consistently across PSROs but who still are somewhat 
problematic when calculating use rates are "Medicare 
crossovers." These are patients whose primary payor 
is Medicare, but who also are medicaid eligible. Many 
Medicaid patients in the OAA, AB, and APTD groups 
are "crossovers". When these patients are admitted 
to the hospital they appear on the PHDDS as Medicare 
patients, since Medicare is the primary payor, even 
though Medicaid may eventually be paying a portion 
of their bill. Because of this coding convention, It must 
be noted that the PHDDS does not really provide data 



63 



from which a total "use" rate can be calculated for 
the Medicaid population. Instead, the PHDDS can pro- 
vide data on rates of those for whom Medicaid is the 
primary payor. Appropriate calculation of use rates 
necessitated removing the "crossover" eligibles from 
the denominator since they are "missing" from the 
numerator as well. 

2.4.3.2.5 Patients Who Die in Hospital 

One further category of patients which requires 
special treatment in a utilization analysis is those who 
die in the hospital. Since an objective of this study is 
to capture total hospital use, dead discharges are 
included in the analysis of discharge rates, if they 
were to be excluded, apparent changes in utilization 
could arise as a result of a change in the proportion 
of dead to total discharges. 

Days of care and discharge rates were calculated in 
this study to include both live and dead discharges, 
while changes in the two component populations were 
shown. Average length of stay, on the other hand, 
cannot be impacted on by the PSRO for dead dis- 
charges. ALOS was thus calculated only for live 
discharges. 

2.4.3.3 MEDICAID ELIGIBILITY (DENOMINATOR) 
ISSUES 

Estimation of the eligible population for use in 
Medicaid rate analysis entails numerous methodologi- 
cal problems and data limitations. In this section the 
issues and problems inherent in Medicaid eligibility 
data are discussed, and the adjustments made to the 
data from each state and PSRO to produce study rate 
estimates are described. 

2.4.3.3.1 Data Variability Among States 

Several recurring problems which appeared as 
eligibility data were requested from the Medicaid state 
agencies. They are listed here: 

• Program categories and eligibility criteria vary 
considerably from state to state, with some Medi- 
caid eligibles covered under Federal matching 
programs and other eligibles participating in 
"state-only" programs. As a result, it was often 
difficult to distinguish Federal from "state-only" 
groups or to establish comparable categories 
among states. 

• Retention of data, either on-line or via magnetic 
tape, varied among the states. Also, the format 
of archived data sometimes changed within 
PSROs during the study period. 

• Specificity of available hard-copy data by county, 
program, age, and sex varied widely among the 
States. In ail cases, the State's own computer data 
processing requests took priority over special 
requests for detailed eligibility data. 

• Since it was not possible to obtain counts of 
Medicaid crossovers by program, age, and sex 



from existing State reports, a crossover estimate 
was made from existing Federal data, and counts 
were adjusted based on the estimate. 

2.4.3.3.2 Methods of Counting Eligibles 

Participation of the Eligibility Pool: A major difficulty 
in interpreting the reported counts of Medicaid eligi- 
bles is that the number of individuals on the 
eligibility file generally represents only a portion of 
the people in the state who are potentially eligible 
for Medicaid at any point in time. These potential 
eligibles include both the people who are actually 
receiving assistance (formally on the eligibility file) 
and those who would be qualified if they were to apply. 
Identification of the total "pool" of potential eligibles is 
extremely difficult. The ratio of the number of persons 
on the eligibility file to the potentially eligible popu- 
lation is the participation rate. The participation rate 
for the eligible population changes over time, signifi- 
cantly affecting the volume and the utilization rate of 
services delivered under the Medicaid program. Con- 
founding the problem of changes in the participation 
rate are increases or decreases in the size of the pool 
of eligibles as a result of changes in the state eligi- 
bility requirements. As discussed by Boland,'*^ the 
changes in the size of the pool and the participation 
rate are likely to interact with other factors to affect 
the utilization rate observed for the Medicaid 
population. 

For this study, no information could be obtained on 
the total size of the eligible pool or the changes that 
may have occurred in the participation rate over the 
time period in the participating PSROs. 

Turnover of the Eligible Population: Three alternate 
methods of counting Medicaid eligibles were 
identified: 

• Number of persons who were ever eligible dur- 
ing a specific period (e.g., a quarter). 

• Number of "person year equivalents" of eligibility. 

• Number of persons eligible at a specific point in 
time. 

Because of high turnover among Medicaid eligibles, 
these three approaches may result in substantially 
different estimates of the Medicaid population to be 
used in the denominator of a rate calculation. The first 
method, persons ever eligible, results in the largest 
estimate of the population in that it counts all indi- 
viduals on the file during a specific period regardless 
of how long they remained eligible. The second method 
summarizes monthly eligibility counts into Medicaid 
eligible full-year equivalents, thereby adjusting for the 
frequent turnover often observed in the Medicaid pro- 
gram. This method produces the smallest population 
estimates. The third method, eligibles at a point in 



' B. Boland, 'The Family, Poverty, and Welfare Programs: 
Factors Influencing Family Instability," Paper No. 12 in 
Studies in Public Welfare, Joint Economic Committee, 
U.S. Congress, 1973. 



64 



time, simply counts the number of eligibles as of a 
specific month. This was the method selected for this 
study. It should be noted that in one PSRO the count 
of the number of persons ever eligible for a year was 
almost double that of the monthly count, reflecting ap- 
proximately a 50 percent turnover rate in the Medicaid 
eligible population. 

2.4.3.3 MISSING DATA 

Among the states, eligibility counts were not avail- 
able for every quarter in the analysis. In order to per- 
mit calculation of rates for each quarter, straight line 
interpolation was used to provide estimates for the 
missing quarters. 

2.4.4 Methodology for Case Study Analysis 

2.4.4.1 UTILIZATION MEASURES STUDIED 

Given the issues described above, each of the four 
study PSRO areas was treated as an independent case 
study in which the emphasis was to describe changes 
in hospital utilization by Medicaid beneficiaries over 
the period of analysis. The three principal measures 
used to study utilization patterns for Medicaid are the 
same measures used in earlier Medicare studies: 

• Days of Care Rate (DOC): the number of hospital 
days per 1,000 Medicaid eligibles. 

• Discharge Rate (DISC): the number of discharges 
per 1,000 Medicaid eligibles. 

• Average Length of Stay (ALOS): the number of 
hospital days per hospital discharge. 

For the four PSROs, these measures were calculated 
for Medicaid discharges for each quarter during the 
study period. Rates were calculated using the formula 
presented in Section 2.4 3.2.5. The ALOS was calcu- 
lated from discharges and days of care data. Because 
of the difficulty in estimating "average" eligibility for 
an entire year, annual rates were not calculated. 

2.4.2.2 ADJUSTMENTS TO UTILIZATION MEASURES 

In the analysis of hospital utilization, three adjust- 
ments are often appropriate to standardize the crude 
rate measures for longitudinal comparisons. An age- 
sex adjustment is useful to control for any changes 
that might have occurred in the age-sex structure of 
the Medicaid eligible population over the three-year 
time period included in the study. An adjustment that 
accounts for changes in the case-mix of the patient 
population over time may also be appropriate. Finally, 
a third adjustment to eliminate changes in the mix 
of beneficiaries by eligibility category may be de- 
sirable. These adjusments were not applied to data 
from the four case studies developed in this analysis. 
However, changes in the age-sex, case-mix, and pro- 
gram category distributions for the PSRO case study 
areas were not large over the observation period. Thus, 
the trend data presented are not thought to be heavily 



influenced by lack of adjustment for these distri- 
butions."^ 

2.4.4.3 HYPOTHESES 

Two principal hypotheses were examined for each 
of the utilization measures. 

Days of Care Rate 

Hj: The overall days of care rate in the PSRO did 
not change over the time period 1976 to 1978. 

H2'. The days of care rate for the relatively most 
common diagnoses in the PSRO decreased 
over the time period 1976 to 1978. 

Discharge Rate 

H3: The overall discharge rate in the PSRO did not 
change over the time period 1976 to 1978. 

Hi'. The discharge rate for the relatively most com- 
mon diagnoses in the PSRO decreased over the 
time period 1976 to 1978. 

As in the days of care rate analysis, it was not ex- 
pected that the overall DISC rate would decline 
significantly over the time period for the PSROs in- 
cluded in the study. However, the DISC rate for 
specific diagnoses was expected to decline for those 
diagnoses on which the PSROs focused their review 
efforts. 

H5: The overall average length of stay in the PSRO 

did not change over the time period 1976 to 

1978. 
Hg: The average length of stay for the relatively 

most common diagnoses in the PSRO decreased 

over the time period 1976 to 1978. 

It was expected that changes in ALOS would be 
small and probably not statistically significant for all 
of the PSRO cases (even for age-sex specific groups). 
It was expected, however, that PSRO review would 
affect the shape of the LOS distribution by reducing 
the number of short stays. The distribution was also 
expected to change by reducing the percentage of 
long stays as a result of concurrent review activity 
and discharging or transferring of patients from acute 
care hospitals on a more timely basis. 

2.4.4.4 ANALYSES AND TECHNIQUES 

2.4.4.4.1 Analyses 

This section discusses the technique used to 
analyze the three utilization measures, DISC, DOC, 
and ALOS. Because the proportion of obstetrical dis- 
charges was observed to increase over time, overall 
rates were analyzed including and excluding obstetri- 
cal discharges. Additionally, rates were prepared for 



This judgment is in part confirmed by the Case-Mix 
Adjustment Study (Section 2.5) which demonstrates 
relatively small changes in case-mix, over the observation 
period, for hospitalized Medicaid beneficiaries. 



65 



22 selected diagnoses and four surgical status 
categories. 

The 22 diagnoses were selected primarily because 
they occurred with relatively high frequency for case 
study area Medicaid patients. A preliminary analysis 
showed that each of the listed diagnoses represented 
more than one percent of all discharges (excluding OB 
and newborns) in at least two of the PSRO case study 
areas. The diagnoses selected for the diagnosis-specific 
analysis are presented in Table 34. The corresponding 
code list for H-ICDA-2 and ICDA-8 are necessary be- 
cause PHDDS data contained both types of coding 
during the study period. 

The reason for this diagnosis-specific analysis is 
that although no change (or no statistically significant 
change) may be observed across the utilization ex- 
perience of all patients in a PSRO, changes may be 
observed in specific diagnoses as a result of changing 
physician and hospital practice patterns and, perhaps, 
as a result of focused PSRO review activities. Caution 
must be exercised in interpreting the diagnosis-specific 
analyses, especially when speculating about PSRO 
impact. Re-abstracting studies have indicated that 
anywhere from 20 to 40 percent of all diagnostic cod- 
ing is inaccurate. Surgical procedure coding is 
somewhat more reliable. In part this may be due 
to the fact that the medical record often contains 
pathological confirmation of surgical diagnosis and the 
abstractor has access to these data, while for medi- 



cal diagnoses the abstractor must complete diagnostic 
coding before the physician's discharge summary is 
prepared. Despite the fact that some proportion of all 
diagnoses are mis-coded, the diagnostic data pre- 
sented provide baseline data about the case-mix of 
patients in each of the PSROs in the study. 

The analysis of surgical status categories identified 
four classes as follows: 8« 

• Non-Surgical 

• Surgical 

• Obstetrical 

• Endoscopy and other diagnostic procedures. 
Obstetrical and endoscopy patient discharges were 
counted as separate, specific categories for these 
reasons: 

• The special nature of these procedures 

• Current debate in the medical community as to 
whether or not these procedures represent sur- 
gical intervention 

• Inconsistency among PSROs and even within the 
same PSRO over time in coding these procedures 
on the discharge abstract. 

An additional analysis of rates and ALOS for age- 
sex specific categories was conducted since patterns 

** Lists of procedures that were classified as non-surgical 
discharges, obstetric discharges, and endoscopic 
discharges are included in the contractor's report (O'Brien 
et al., op. cit.). 



TABLE 34 
Diagnosis Categories by Code: H-ICDA-2 and ICDA-8 



Category 



H-ICDA-2 



ICDA-18 



1 


Obstetrical 


2 


Intestinal Infectious Disease 


3 


Diabetes 


4 


Schizophrenia 


5 


Affective Psychoses 


6 


Neuroses 


7 


Other Mental Disorders 


8 


Otitis Media 


9 


Acute URI, except Influenza 


10 


Pneumonia 


11 


Acute Bronchitis & Bronchiolitis 


12 


Bronchitis, Chronic/Unspecified 


13 


Asthma 


14 


Hypertrophy of T&A 


15 


Inguinal Hernia 


16. 


Disease-Pelvic/Parametrium 


17 


Menstrual Disorders 


18 


Cellulitis 


19 


Abdominal Pain 


20 


Convulsions 


21 


Concussion 


22 


Sterilization 


23 


All others 



631-678 


630-678 


001-009 


000-009 


250 


250 


306 


295 


308 


296 


310 


300 


290-305, 307, 


290-4, 297-9, 


309, 311-318 


301-315 


381 


381 


460-465 


460-465 


480-486 


480-486 


489 


466 


490-491 


490-491 


493 


493 


500 


500 


550 


550 


616 


616 


626 


626 


682 


682 


780.0 


785.5 


770.3 


780.2 


850 


850 


Y13.0 


Y09.0 


All other 


All other 



66 



of hospital use are known to vary considerably by age 
and sex. In particular, the individual age-sex classes 
are somewhat of a proxy for program categories. For 
example: 

• Most of the population under 21 are dependent 
children covered under AFDC: some females 
17-20 are mothers of AFDC children. 

• The under 21 population also contains some blind 
and disabled of both sexes. 

• Most middle-aged male beneficiaries are blind or 
disabled while middle-aged women are divided 
between AFDC and disabled or blind. 

• Most of the eligibles over 65 are OAA although 
some are blind or disabled or adult caretakers of 
dependent children through AFDC. 

2.4.4.4.2 Techniques 

The two primary analytic techniques used in this 
study were trend analysis and comparisons of matched 
quarters within PSRO case study areas. More precise 
evaluative techniques could not be employed for 
PSRO impact analyses since no pre-post or non-PSRO 
area comparison data were available. 

Trend Analyses 

A descriptive analysis of calculated rates and ALOS 
for the study time period was performed for each case 
study PSRO. Rates were calculated for each quarter, 
thus providing twelve observations for the PSROs with 
three years of data and eight observations for the 
PSROs with two years of data. The slope of the trend 
lines was calculated using a least squares technique. 
The statistical significance of the slope was examined 
using Student's t-test values. 

Utilization measures were calculated for Medicaid 
for selected diagnoses. As indicated, hospital utiliza- 
tion was also analyzed separately for medical and 
surgical patients. Slopes of the medical and surgical 
days of care rate trend lines were calculated. The 
graphic displays of the plotted data were visually in- 
spected for apparent seasonal variation and abrupt 
changes. 

Matched Quarter Analyses 

Utilization measures were also analyzed matching 
the same quarters from each study year. The days 
of care rate for the first quarter of the calendar year 
was compared across the time period for each year 
included in the study of each PSRO. The process was 
reported for the second, third, and fourth quarters as 
well as for the total year. The precent changes in the 
days of care rate were examined to determine if any 
consistent pattern of increase or decrease in the rate 
was present. This technique in effect adjusts for sea- 
sonal variations. 

2.4.5 Case Study Findings 

2.4.5.1 INTRODUCTION 

This section presents descriptive utilization data 
and a discussion of important findings. Background 



comparative statistics are presented for the case study 
areas in Table 35. The four PSROs in this analysis are 
all mature PSROs which have been in operation since 
the beginning of the PSRO program. Three of them 
were in existence and performing reviews before re- 
ceiving funding as a PSRO. Three of the four PSROs 
are located in the Western region of the United States, 
which has historically experienced the lowest hospital 
utilization levels in the country. Thus, these four 
PSRO areas do not represent a national sample. 

For two primary reasons, rates should not be com- 
pared among the PSROs. First, out-migration is high 
in two areas (up to 20-25% in PSRO 1 and PSRO 3) 
causing utilization rates to be artificially low. Second, 
the composition of the Medicaid population is very 
different in PSRO 4 (only 58.7% of eligibles were 
AFDC eligibles). This may be partially responsible for 
the high use rates observed in PSRO 4. A summary of 
case study results follows the separate presentation 
of each of the four case studies. 

2.4.5.2 CASE STUDY ANALYSES 

2.4.5.2.1 PSRO 1 

Descriptive utilization and eligibility data by quarter 
are presented in Table 36 for PSRO 1. Trend analyses 
showed a slight decline in days of care rate between 
1977 and 1978. Matched quarter analysis showed that 
increases in the second and fourth quarters were offset 
by decreases in the first and third quarters to produce 
the overall decrease. Modest declines occurred for 
ALOS while discharge rates increased slightly. 

For diagnostic specific analyses the only significant 
trends in days of care rates were an increase for 
convulsions and a decrease for acute URI. Diabetes 
had a significant decline in ALOS. The surgical status 
analysis showed that decreases in days of care rates 
for surgical, non-surgical, and endoscopic discharge 
were paralleled by decreases in ALOS. 

2.4.5.2.2 PSRO 2 

For PSRO 2, basic quarterly utilization and eligibility 
data are displayed in Table 37. A nonsignificant in- 
crease in days of care rate was observed from 1976- 
1978. In the matched quarter analysis, the largest in- 
creases in days of care rate were observed to occur 
during the second and third quarters. Days of care 
rate increases were accompanied by increases in the 
discharge rate and ALOS. 

Mixed results were observed in the diagnostic- 
specific studies but there were significant increases 
in the days of care rate for Affective Psychoses and 
the ALOS for Other Mental Disorders. The non-surgi- 
cal, surgical, and obstetric groups showed increases in 
all three utilization measures. Declines occurred in all 
three measures for endoscopic discharges. 

2.4.5.2.3 PSRO 3 

Since the collection of obstetrical data was incon- 
sistent in PSRO 3 during the study period, these dis- 



67 



TABLE 35 
Description of Case Study Areas 









Case Study Areas 






Characteristic 


PSRO 1 


PSRO 2 


PSRO 3 


PSRO 4 




PSRO Review 










1. 


PSRO awarded conditional 
status 


June 1975 


June 1974 


June 1974 


June 1975 


2. 


Type of PSRO review 


Delegated 


Non-Delegated 


Mixed 


Primarily non- 
delegated 


3. 


Size of PSRO 


County 


State 


County 


State 


4. 


Start of PSRO data processing 
Medical Supply Variables 


Fall 1976 


June 1974 


Fall 1974 


Late 1975 


1. 


Number of hospitals 


7 


40 


14 


67 


2. 


Beds per 1 ,000 residents 


4.5 


3.6 


8.0 


5.7 


3. 


Nursing care beds per 1,000 
aged residents 


36.7 


42.5 


48.2 


61 


4. 


Physicians per 1,000 residents 
Medical Demand Variables 


1.8 


1.6 


3.6 


1.2 


1. 


Number of Medicare enrollees 


51,500 


105,500 


81,000 


79,000 


2. 


Number of Medicaid eligibles 


47,500 


43,500 


48,500 


30,000 


3. 


AFCD as percent of all 
Medicaid eligibles 

Demographics 


81.6 


75.9 


78.1 


57.7 


1. 


Census Region 


Northeast 


West 


West 


West 


2. 


Level of out-migration for 
hospital care 


High 


Low 


High 


Low 


3. 


Number of Title XVIII 
discharges 


18,400 


35,400 


30,100 


35,900 


4. 


Number of Title XIX discharges 


9,900 


12,400 


9,700 


10,700 


5. 


Relative Urbanization 


High 


Mostly low, 
high in one 
area 


High 


Low 



TABLE 36 

PSRO 1: Quarterly Counts of l\iedicaid Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, 1977-1978* 



Year 



Qtr 



Number of 
Eligibles 



Discharges 



Daye of 
Care 



ALOS* 



Discharge 
Rate 



Days of 
Care Rate 



77 


1 


43621 


2102 


13737 


6.5 


48.2 


314.9 


77 


2 


43538 


2076 


12486 


6.0 


47.7 


286.8 


77 


3 


43450 


2219 


14048 


6.3 


51.1 


323.3 


77 


4 


43263 


2045 


13127 


6.3 


47.3 


303.4 


78 


1 


43330 


2043 


13452 


6.5 


47.1 


310.5 


78 


2 


43244 


2132 


12688 


5.9 


49.3 


293.4 


78 


3 


43137 


2115 


12304 


5.8 


49.0 


285.2 


78 


4 


43333 


2177 


13529 


6.2 


50.2 


312.2 



Newborns Excluded 
Live Discharges Only 



68 



TABLE 37 

PSRO 2: Quarterly Counts of Medicaid Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, ISye-lSyS* 



Year 



Qtr 



Number of 
Eligibles 



Discharges 



Daye of 
Care 



ALOS* 



Discharge 
Rate 



Days of 
Care Rate 



76 


1 


48016 


2779 


15902 


5.7 


57.9 


331.2 


76 


2 


47031 


2417 


13745 


5.7 


51.4 


292.3 


76 


3 


45990 


2606 


14466 


5.5 


56.7 


314.5 


76 


4 


44086 


2450 


15856 


6.4 


55.6 


359.7 


77 


1 


42499 


2583 


15122 


5.7 


60.8 


355.8 


77 


2 


44299 


3013 


16243 


5.4 


68.0 


366.7 


77 


3 


43504 


3121 


16332 


5.2 


71.7 


^ 375.4 


-n 


4 


44494 


3112 


17959 


5.7 


69.9 


403.6 


78 


1 


46503 


2752 


15454 


5.6 


59.2 


332.3 


78 


2 


48512 


2595 


16450 


6.3 


53.5 


339.1 


78 


3 


44971 


2671 


15732 


5.9 


59.4 


349.8 


78 


4 


44838 


2353 


14477 


6.2 


52.5 


322.9 



Newborns Excluded 
Live Discharges Only 



charges were included in counts of total discharges 
and total days of care, but excluded from ail rate and 
length of stay analyses. Table 38 displays pertinent 
data for PSRO 3. The overall trend during the three- 
year study period was a decreasing days of care rate. 
There was no particular seasonal pattern to changes 
in the days of care rate. The discharge rate dropped 
more substantially than the days of care rate, and it 
dropped in all four quarters. In contrast, ALOS in- 
creased slightly. 

For selected diagnoses, the days of care rate rose 
significantly for Other Mental Disorders and fell for 
Asthma, Affective Psychoses, and Menstrual Dis- 
orders. Trends for surgical and non-surgical discharges 
moved in opposite directions. All three measures in- 
creased for surgical patients. Endoscopic discharges 
fell significantly for both days of care rates and ALOS. 

2.4.5.2.4 PSRO 4 

Table 39 displays quarterly data elements for this 
PSRO. Basic data indicate seasonal trends — high in 
the first quarter and low in the third quarter — ^for both 
total discharges and total days of care. The days of 
care rate exhibited an overall positive trend over the 
three-year time period. Consistent increases appeared 
in all but the third quarter. The increase in days of 
care rate was caused by rising discharge rates. Essen- 
tially no change was found in ALOS over the three 
years. 

Among the diagnoses. Schizophrenia, Affective 
Psychoses, and Other Mental Disorders all exhibited 
significant increases in days of care rate. For all of 
them, except Menstrual Disorders, discharge rates 
rose. For Non-Surgical, Surgical, and Obstetric dis- 
charges, higher days of care rates were joined by 
higher discharge rates. 



2.4.5.3 SUMMARY OF CASE STUDY FINDINGS 

2.4.5.3.1 Overall Rates 

Aggregate Medicaid utilization counts — total dis- 
charges and total days of care — tended to fluctuate 
substantially from quarter to quarter during the study 
years. Furthermore, no seasonal trend was apparent 
in three of the four PSRO areas. The only exception, 
PSRO 4, exhibited relatively high counts of discharges 
and days of care in the first quarter of each year and 
relatively low counts in the third quarter of each year. 
Since the age/sex, case-mix, and program composi- 
tion remained stable for the study years and in-migra- 
tion patterns did not change significantly, observed 
fluctuations in the utilization counts were probably 
caused by other factors, such as changes or incon- 
sistencies in data reporting, variations in patterns of 
out-migration and inclusion of services for "short- 
term" eligibles. 

Fluctuations in quarterly counts of Medicaid eligi- 
bles were even more striking than the fluctuations in 
utilization. For example, reported counts of Medicaid 
eligibles differed by 11 percent in successive quarters 
for PSRO 4. Although fluctuations in other areas were 
not as dramatic, they occurred regularly. Other areas 
may have experienced either inaccuracies or changes 
in reporting. Other causal factors may be changes in 
the definition of eligibility and, as noted above, inclu- 
sion of short-term eligibles. High turnover in some 
eligibility classes is known to exist in PSRO 4. 

Since both the utilizaion counts (numerator data) 
and eligibility totals (denominator data) tended to 
fluctuate, it was reasonable to expect that utilization 
rates would also show substantial fluctuation. Succes- 
sive quarterly changes in both discharge and days of 
care rates exceeded 25 percent in one PSRO and 10 
percent in another PSRO. 



69 



TABLE 38 

PSRO 3: Quarterly Counts of Medicaid Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, 1976-1978* 



Year 



Qtr 



Number of 
Eligibles 



Discharges 



Days of 
Care 



ALOS* 



Discharge 
Rate 



Days of 

Care*** 

Rate 



76 


1 


40678 


1877 


13314 


7.5 


41.2 


310.2 


76 


2 


41731 


1872 


12518 . 


7.1 


39.7 


281.7 


76 


3 


42743 


1951 


13980 


7.7 


40.3 


309.4 


76 


4 


43754 


2034 


13885 


7.2 


41.8 


300.8 


77 


1 


44894 


1967 


13342 


7.1 


39.6 


282.8 


77 


2 


44814 


2301 


14694 


7.2 


41.3 


297.9 


77 


3 


44729 


2298 


15338 


7.5 


42.2 


316.3 


77 


4 


44695 


2217 


14397 


7.4 


39.6 


294.0 


78 


1 


44656 


2235 


13986 


7.0 


40.4 


284.0 


78 


2 


43755 


2129 


14153 


7.5 


39.0 


294.1 


78 


3 


42847 


2041 


13512 


7.7 


36.8 


284.2 


78 


4 


41937 


1955 


13261 


7.8 


37.5 


290.8 



Newborns Excluded 

Live Discharges only, excluding Obstetric discharges 

Excluding Obstetric discharges 



TABLE 39 

PSRO 4: Quarterly Counts of Medicaid Eligibles, Discharges, Days of Care, 
ALOS, and Rates per Thousand Eligibles, 1976-1978 



Year 



Qtr 



Number of 
Eligibles 



Discharges 



Days of 
Care 



ALOS' 



Discharge 
Rate 



Days of 
Care Rate 



76 


1 


23647 


2558 


12792 


4.9 


108.2 


541.0 


76 


2 


25223 


2429 


12860 


5.3 


96.3 


509.9 


76 


3 


24538 


2123 


10512 


4.9 


86.5 


428.4 


76 


4 


23788 


2132 


10726 


4.9 


89.6 


450.9 


77 


1 


23037 


2623 


13178 


4.9 


113.9 


572.0 


77 


2 


22283 


2335 


11643 


4.9 


104.8 


522.5 


77 


3 


21531 


2155 


10912 


5.0 


100.1 


506.8 


77 


4 


20808 


2360 


12729 


6.3 


113.4 


611.7 


78 


1 


21895 


2552 


13164 


5.1 


116.6 


601.2 


78 


2 


19788 


2332 


11608 


4.9 


117.8 


586.6 


78 


3 


21963 


1881 


9678 


5.1 


85.6 


440.7 


78 


4 


21390 


2044 


10971 


5.2 


95.6 


512.9 



Newborns Excluded 
Live Discharges Only 



The net effect of these variations in quarterly rates 
was mixed. Two PSRO areas had overall declining pat- 
terns in DOC for the study period while the other two 
PSRO areas had increasing patterns. For the two 
PSROs that experienced declines in DOC, the decline 
was driven by a reduction in the DISC (ALOS in- 
creased) in one area and by reductions in the ALOS 
(DISC increased slightly) in the other area. In the two 
PSRO areas where DOC increased, DISC also in- 



creased. Changes in DOC appear to be primarily re- 
lated to changes in DISC rather than changes in 
ALOS. A summary of these and other important find- 
ings is included in Table 40. 

Preliminary analyses of age-specific rates indicate 
that wide differences exist in DOC, DISC, and ALOS 
between males and females for certain age classes. 
DOC, DISC, and ALOS were consistently higher for 
males than for females in the middle age groups. 



70 



Some of the observed differences may have occurred 
because of the relatively high frequency of pregnan- 
cies and obstetrical discharges for females and dis- 
ability for males in those age groups. 

2.4.5.3.2 Selected Diagnoses 

For the 22 selected diagnoses, DOC changes also 
showed mixed results across the four PSRO areas. In 
the two PSRO areas that experienced overall declines 
in DOC, approximately half of the diagnoses showed 
increasing DOC and the other half showed decreasing 
DOC. In the PSROs showing overall increases in DOC, 
thirteen diagnoses showed declining DOC and nine 
showed increasing DOC. 

It is interesting to note that several diagnoses 
groups showed consistent patterns in DOC. Six diag- 
noses (Otitis Media, Pneumonia, Acute URI, Acute 
Bronchitis, Chronic Bronchitis, and Menstrual Dis- 
orders) had declining DOC in all four PSRO areas. 
Those declines were driven by consistent declines in 
the DISC for all but one of those diagnoses. Otitis 
Media. The overall trend in ALOS was downward but 
there were some increases in ALOS for these six 
diagnoses. 

Consistent increases in DOC were observed for five 
diagnoses (Intestinal Infectious Disease, Obstetric, 
Schizophrenia, Neuroses, and Other Mental Disorders). 
Here, increases in DOC were frequently accompanied 
by increases in DISC. Frequently, for the three men- 



TABLE 40 
Findings of Quarterly Trend Analysis, 1976-1978 





PSRO Trends 




1976-78* 


All Patients: 




Days of Care/ 1,000 


2 Down, 2 Up 


Discharges/1,000 


1 Down, 3 Up 


Average Length of Stay 


1 Down, 2 Up, 1 Same 


Non-Surgical DOC/1,000 


2 Down, 2 Up 


Surgical DOC/1,000 


1 Down, 3 Up 


Endoscopic DOC/1,000 


4 Down 


Specific Diagnoses DOC/1,000: 




Otitis Media 


4 Down 


Pneumonia 


4 Down 


Acute URI 


4 Down 


Acute Bronchitis 


4 Down 


Chronic Bronchitis 


4 Down 


Menstrual Disorder 


4 Down 


Intestinal Infectious 




Disease 


4 Up 


Schizophrenia 


4 Up 


Neuroses 


4 Up 


Other Mental Disorders 


4 Up 


Obstetrical 


4 Up 



* Trends are based on the years 1977 and 1978 only in 
PSRO 1. 



tally related illnesses (Schizophrenia, Neuroses, and 
other Mental Disorders) increased ALOS and DISC 
contributed to increases in DOC. In the future, analyses 
will focus on the components of the Medicaid popula- 
tion where increasing utilization for these mentally 
related disorders has occurred. 

2.4.5.3.3 Surgical Status Groups 

The surgical status analysis showed mixed results 
across the four PSRO areas for non-surgical dis- 
charges, but one observation was possible. Increases 
in the DOC for non-surgical patients were accompanied 
by increases in DISC, while decreases in the DOC 
were accompanied by decreases in the ALOS. Gen- 
erally, DOC and DISC for surgical patients rose, but 
no consistent pattern emerged. Finally, patients having 
endoscopic and other diagnostic procedures experi- 
enced declines in both DOC and DISC for all four 
areas. 

2.4.5.3.4 Discussion of Case Study Findings 

In the four PSROs under study, DOC increased in 
two single State PSROs and decreased in the two 
countywide PSROs. In no case was the rate change 
statistically significant. 

The relationship between DOC and DISC seemed to 
be the most direct. That is, when DOC increased, it 
was usually associated with an upward trend in DISC- 
The reverse was also true. This finding is consistent 
with results of prior PSRO evaluation studies and im- 
plies that PSRO review programs which have tradi- 
tionally emphasized length of stay reduction may have 
potentially more impact in reducing admissions to hos- 
pitals, particularly for Medicaid patients. 

The analysis by diagnosis group revealed the same 
pattern. Declines in the DOC were generally associated 
with decline in DISC for a given diagnosis. In two 
PSROs there appeared to be a tendency for the ALOS 
to increase, even when the DOC decreased, for several 
diagnosis groups. The general variation in use rates, 
observed in all areas, may be a result of the interplay 
of one or more factors that include data artifacts, re- 
porting inconsistencies, changing migration patterns, 
transient eligibility, Medicaid program changes, and 
lack of specificity among data elements. 

2.4.5.4 POTENTIAL OF THE DATA BASES FOR 
MEDICAID RATE ANALYSES 

Rate analyses conducted on PHDDS data in this 
study were limited by the problems of defining ap- 
propriate eligibility populations-at-risk and selecting the 
appropriate "unit" of measure of eligibility. The link- 
age of PHDDS utilization data (numerators) with Medi- 
caid eligibility data (denominators) is hampered by 
data problems discussed earlier such as: 

• inclusion of "state only" patients, 

• adjustment for Medicaid "crossovers", 

• inclusion of "spend down" eligibles, 



71 



• disposition of retroactive or indeterminate eli- 
gibles, and 

• inconsistent reporting for newborn and obstetrical 
cases. 

Tfie most serious problems relate to estimating the 
population-at-risk but other problems arise when using 
the PHDDS as the source of utilization data. Problems 
with eligibility data can be reduced as special requests 
for detailed eligibility data are satisfied by individual 
state agencies. However, since the state agencies are 
the only source of eligibility data, limiting problems 
remain. 

On the other hand, PHDDS data are of limited use 
for rate analyses within national PSRO evaluative 
efforts, because they exist only for "active" PSRO areas 
and they are not broken out by program category. 
Moreover, area comparisons are not possible because 
patient migration data are not generally available. In 
the absence of other good sources of Medicaid utiliza- 
tion data, it is important to document under what cir- 
cumstances the PHDDS may provide usable informa- 
tion. The following section outlines steps that can be 
taken to improve the utility of PHDDS data for Medi- 
caid analysis and presents other alternatives. 

2.4.6 Recommendations 

The recommendations have been broken into two 
major classes: 

• Recommendations for further analysis based on 
PHDDS data, and 

• Recommendations for analyses based on State 
agency Medicaid claims. 

These recommendations represent options along a 
continuum ranging from minor improvements to a 
total reconstruction of Federal Medicaid reporting. 

2.4.6.1 FURTHER ANALYSIS WITH RESPECT TO 
PHDDS DATA 

Recognizing the limitations inherent to PHDDS data, 
utility of the PHDDS could be enhanced if data re- 
garding program category, dual eligibility, and eligi- 
bility for "state only" programs could be routinely re- 
ported as a part of an "augmented" PHDDS data set. 
Specific recommendations are outlined below: 

Exploratory Efforts 

• Survey PSROs to determine the information coded 
on the expanded PHDDS maintained by their data 
processors, with emphasis on Medicaid identifi- 
cation number, program identification information, 
and patient county of residence. 

• Investigate the specific content of the Medicaid 
identification number, inclusion of any program 
identifier code, "crossover" indicator and avail- 
ability of detailed State eligibility data. 

• Analyze Medicaid utilization by categorical pro- 
gram of eligibility, using PHDDS files obtained di- 
rectly from the PSRO, augmented by the part of 



the patient's Medicaid identification number that 
specifies program category. 

• Investigate more thoroughly the appropriateness 
of alternative methods (such as number of eli- 
gibles at a point in time, persons ever eligible, 
and person-years of eligibility) to count Medicaid 
eligibles for use as the denominator in rate 
analysis. Special emphasis should be given to the 
relationship of program category eligibility turn- 
over rates and participation rate in the eligibility 
estimation procedure. Approaches for calculating 
the eligible population for a month, quarter, and 
year should be examined. 

• To the extent possible, examine the migration of 
Medicaid patients across PSRO boundaries, both 
within and among states, using PHDDS data from 
contiguous PSROs so that adjustments could be 
made to eliminate distortions caused by out- 
migration. 

Management actions 

• Implement procedures among the PHDDS data 
processors to improve their ability to provide re- 
search data files. 

• Implement a consistent policy of including or ex- 
cluding patients with retroactive eligibiltiy or in- 
determinate eligibility status. 

• Enforce current PHDDS data policy regarding the 
inclusion of newborn and obstetrical discharges. 

• Investigate the feasibility of adding a Medicaid 
program eligibility category to the PHDDS as a re- 
quired data element to allow program-specific 
analyses of utilization. This may not be possible 
where the Medicaid identification number is the 
same as the Social Security number. 

• Recommend that HGFA initiate a Memo of Under- 
standing (MOU) with Medicaid State Agencies re- 
garding standardized reporting of Medicaid eligi- 
bility data by age, sex, and program category, in- 
cluding estimates of Medicare "crossovers." The 
recommendation should include the form and fre- 
quency of reporting. 

2.4.6.2 ANALYSIS BASED ON MEDICAID CLAIMS 
DATA 

While the efforts described above would enhance 
PHDDS Medicaid analyses, many of the limitations of 
PHDDS data described in earlier sections could not be 
overcome. Only after all areas of the country are fully 
implemented, will PHDDS data be available to com- 
pletely count out-migration. No data will ever be re- 
ported through the PHDDS for non-PSRO areas or pre- 
PSRO activity. Since PHDDS data and State Medicaid 
eligibility are collected by dissimilar data systems in 
different political jurisdictions, certain limitations will 
exist as researchers attempt to link PHDDS numera- 
tors and State eligibility denominators. Given all of the 
problems related to the use of PHDDS data, the use 
of State Medicaid claims as a source of utilization 
data would have several advantages as outlined below: 



72 



1. Data could be requested for States that contain 
both active and inactive PSROs. 

2. Data for some states could be gathered to in- 
clude both pre-PSRO and post-PSRO utilization. 

3. Specific rates by program category could be 
more rapidly developed and defined more care- 
fully with State data than for PHDDS data. Also, 
linkage of claims and eligibility information is 
more consistent when one data source provides 
both sets of data. 

4. Patients entitled for "state only" programs and 
patients having dual eligibility under both Medi- 
care and Medicaid could be more easily identi- 
fied and isolated in many areas. 

Specific recommendations for the analysis related to 
Medicaid claims are listed below: 

Exploratory efforts 

• Survey Medicaid State agencies to determine the 
feasibility of obtaining unit records for selected 
data elements from Medicaid claims history files 
and Medicaid eligibility files to build a national 
Medicaid data base. 

• For areas where State Medicaid data are avail- 
able, acquire data, calculate program category 
specific rates for various eligibility definitions, and 
compare the analyses to PHDDS analyses wher- 
ever possible. 

• Extract specific claims and eligibility unit records 
to expand on the alternative methods of rate 
calculation by studying various "cohorts" of 
Medicaid beneficiaries. 

• Study Medicaid patient migration patterns and 
their changes over time. 

• Study the levels and patterns of Medicaid utiliza- 
tion for Medicare "crossovers," "state only" eli- 
gibles, and estimate utilization for "spend-down" 
patients. 

Summary 

In summation, there is no basis for generalizing the 
results of the case study analyses to estimate national 
trends or to predict the activity of any "similar" PSRO. 
Because Medicaid is a conglomerate of distinctive 
State programs, each having unique data systems, 
eligibility requirements, extent of coverage and vary- 
ing benefit payment structures, comparability of rate 
analyses across states, PSROs or any other set of 
geographic boundaries will be difficult to accomplish. 
Within each State, the composition of the Medicaid 
population with respect to age, sex, and race is con- 
stantly changing. Since individual State programs un- 
dergo constant fluctuation of coverage and benefit 
structure within the State, comparability among areas 
at one point in time may not imply comparability at 
another point in time. Moreover, longitudinal data for 
a single area may contain significant discontinuities. 

This study represents a beginning — laying the 
groundworic for the comparison of Medicaid utilization 



among the States and for the evaluation of PSRO im- 
pact of Medicaid. A lot of painstaking effort to gather 
data, define methodology, and coordinate research ac- 
tivities remains before a final assessment of PSRO im- 
pact can be determined. 



2.5 Case-Mix Adjusted Length of Stay 
Analysis (Case-Mix Adjustment Study) 

2.5.1 Objectives 

The use of aggregate utilization data to measure 
changes in ALOS over time, or to compare ALOS dif- 
ferences across PSROs, implicitly assumes that the 
mix of patients hospitalized within a given PSRO area 
either does not vary or, if it does, that those variations 
have no impact upon ALOS. 

The validity of these assumptions is open to some 
doubt. To the degree that PSRO review activity re- 
duces unnecessary admissions it is likely to result in 
changes in the patient population, the medical condi- 
tions for which they are treated, and the surgical pro- 
cedures which they receive. These changes, in and of 
themselves, affect ALOS. The existence of those 
changes thus complicates the issue of comparing 
ALOS over time. Moreover, if the changes which take 
place in one area differ from those which take place 
in others, the ability to make meaningful area compari- 
sons may also be restricted. 

The PHDDS staging study reported in the 1978 
PSRO evaluation attempted to address this issue by 
examining changes in ALOS and in days of care on a 
diagnosis-by-diagnosis basis and by controlling for 
severity of illness (i.e., single vs. multiple diagnosis, 
with or without complications, etc.). Although the study 
produced useful results, those results were limited by 
two factors: 

a. The lack of a single, composite index of utiliza- 
tion which could serve as a dependent variable 
for testing hypotheses concerning the effec- 
tiveness of PSRO review activity, and 

b. The absence of a mechanism for taking into ac- 
count differences in the characteristics of the 
treated, particularly their age. 

The methods described in this section are designed 
to address the foregoing limitations, while retaining 
the diagnosis-specific and severity-specific focus re- 
flected in last year's analysis. In particular, these 
methods are designed to permit one to distinguish 
between changes in length of stay which are 
attributable to true reductions in ALOS and those 
which are essentially a consequence of shifts in the 
severity and complexity of the cases treated. 

Two techniques are presented. The first is a method 
for calculating a case-mix adjusted length of stay 
index, for individual PSROs or for groups of PSROs. 
Two variants of that method are presented: the first 
is useful for longitudinal examinations of performance, 
the second for cross-sectional (i.e., across PSRO) com- 



73 



parisons. The second technique involves the deriva- 
tion of an index of studying variations in case-mix 
composition; again, two variants are presented, one 
for longitudinal and the other for cross-sectional 
comparisons. 

The indices developed are then applied to a selected 
group of twelve PSROs for which suitable patient-spe- 
cific data were available. Findings are presented for 
each of the PSROs separately, for all twelve jointly, 
and are further subdivided by beneficiary group (Medi- 
caid vs. Medicare). 

2.5.2 Data 

The analysis described in this section is based on 
the PSRO Hospital Discharge Data Set (PHDDS), a 
main component of the PSRO Management Informa- 
tion System (PMIS). PHDDS data consist of the Uni- 
form Hospital Discharge Data Set (UHDDS), along with 
additional variables pertinent to the individual PSRO. 
Data are collected at the level of individual patient 
and hospital, and submitted quarterly to HSQB. For 
this study, data were taken from tapes submitted by 
twelve PSROs for each of the eight quarters of 1977 
and 1978. 

Of particular relevance for the present study is the 
fact that PHDDS data include diagnosis and age, so 
that utilization by particular diagnostic and age groups 
can be computed. Information is also included con- 
cerning the use of surgical procedures or their ab- 
sence, and the presence or absence of secondary 
diagnoses. The main data elements available through 
the use of PHDDS are illustrated in Table 41. 



TABLE 41 
PHDDS Data Elements 



1. PSRO ID 

2. Hospital Code Number (3 digit code assigned 

by PSRO) 

3. Patient Information: 

Date of Birth 

Sex 

Race or ethnic group (white, black, other, 

unknown) 
Zip code of residence 

4. Admission/discharge information: 

Date of admission 
Hour of admission 
Date of discharge 
Patient disposition 

5. Information on diagnoses:* 

Principal diagnosis 

First through fourth secondary diagnoses 

6. Information on procedures:* 

Principal procedure and data 
First through third secondary procedures and 
dates 

7. Expected principal source of payment (Medicare, 
Medicaid, Title V, other Government, self-pay, 
workmen's compensation. Blue Cross, commercial 
insurance company, no charge, other) 

* Diagnoses and procedures are coded by ICDA-8 or 
HICDA-2 prior to January 1, 1979 and by ICD-9-CM 
thereafter. 



2.5.3 Methods 

The major premise on which this study is based is 
that the universe of hospitalized individuals can be 
subdivided into discrete cohorts ("cells") according 
to a defined set of characteristics which tend to be 
related to ALOS. This premise has been amply studied 
and confirmed, most notably through the research ac- 
tivities long under development at Yale University 
which led to the interactive computer system termed 
AUTOGRP.83 

There are a number of ways in which patient popu- 
lations might be subdivided. One such way is through 
the use of the diagnosis related groups, or DRGs, 
specifically developed in conjunction with AUTOGRP; 
there are 383 such groups. Another possibility is the 
use of sets of DRGs, called Initial Groups (IG), of 
which there are 83. Because of time constraints, it was 
not possible to include these diagnosis-related features 
in the present study. Instead, the following 24-cell 
classification scheme was used: 



'See, for example, R. Mills et al., "AUTOGRP: An Interactive 
Computer System for the Analysis of Health Care Data," 
Medical Care, July 1976; also, DHEW/HCFA, Health Care 
Financing Research and Demonstration Series, Report 
No. 1, "AUTOGRP Patient Classification Scheme and 
Diagnosis Related Groups," August 1977. 



Variable 1: Patient age — 0to16 

17 to 44 

45 to 64 

65 to 69 

70 to 74 

75 and over 
Variable 2: Single diagnosis vs. multiple diagnoses 
Variable 3: Surgical vs. non-surgical stay 

Regardless of the classification scheme used, the 
mix of patients for a given PSRO during a given period 
of time is completely described by the percentage of 
discharges falling into each defined cell. For the 24- 
cell classification used in this study, that mix is simply 

Pi. P2. • ■ • . P24 

where p, denotes the percentage of discharges falling 
into the I'*' patient cell. These sets of values vary, of 
course, from PSRO to PSRO and, for any given PSRO, 
over time. 

Using the subscript i to denote the i'** patient cohort, 
it is seen that the ALOS experienced by a given PSRO 
may be written as the weighted average of the aver- 
age length of stay for each cohort separately. That is 
to say: 



(1) ALOS = SPiLOSi 



74 



where Pi is the percentage of discharges falling into 
the if' patient cohort and LOSj is the average length 
of stay experienced by that cohort. 

This relationship highlights the fact that changes in 
ALOS may be effected by changes in either Pj ("case 
mix"), LOSj ("average length of stay for a given 
cohort"), or by both jointly. The methods defined in 
this section are designed to permit each of these 
factors to be studied independently of the other. 

2.5.3.1 CASE-MIX ADJUSTED LENGTH OF STAY 

Equation (1) may be used to measure variations in 
length of stay by weighting each of the observed 
length of stay values, LOS;, by a corresponding case- 
mix "norm", p,*. Thus, the case-mix adjusted ALOS 
for a given PSRO may be written: 



(4) 



CM 



(2) 



(AL0S)ADj = 2Pi*L0Si 



where LOS; is the observed average length of stay for 
patient cohort i, and pj* is a case-mix weighting fac- 
tor, or norm, associated with that cohort. 

For longitudinal comparisons, the norm Pj* is chosen 
to be the corresponding case-mix proportion observed 
in some original time period; for cross-sectional com- 
parisons, the norm is chosen to be the mean value of 
all PSROs under study. In the former case, Pi* is 
invariant over time but varies from PSRO to PSRO; 
in the latter case, Pj* applies uniformly to all PSROs 
but varies from one time period to the next. 

To facilitate comparisons, expression (2) may be 
stated as a Laspeyres type index: 



(3) 



CM/ALOS 



SPi^LOSj 
2Pi*L0Si* 



X100 



In this expression, the denominator represents the 
ALOS "norm". This may be either the ALOS applicable 
to the given PSRO in the defined base period (in the 
case of a time series approach), or the "group 
mean" ALOS of all PSROs under study (in the case 
of a cross-sectional analysis). 

The index (3) reflects the percentage by which a 
given PSRO's ALOS would have varied had its case- 
mix proportions remained unchanged. Values greater 
than 100 denote a "true" increase in ALOS, even 
though the observed (raw) ALOS for that PSRO may 
have declined (due, perhaps, to a less severe case- 
mix). Values less than 100 are similarly interpreted: 
they denote a true reduction in ALOS, even though 
the observed value may have gone up. The defined in- 
dex thus provides a case-mix-free measure of the 
ALOS performance of a given PSRO. 

2.5.3.2 CASE-MIX INDEX 

A second useful measure of performance is derived 
by reversing the weights shown in the numerator of 
(3). In this instance, the case-mix proportion serves 
as the observed value, while the ALOS serves as the 
norm. The expression is written: 



2 PiLOSi* 

TpTlos; 



7X100 



Again, the weights (or "norms") might pertain to 
the original base period in the case of a time series 
analysis, or they might represent the group means of 
the PSROs under study in the case of a cross-sectional 
analysis. In either case, the index reflects the per- 
centage by which a given PSRO's ALOS wou/d have 
varied as a result of case-mix changes alone, i.e., if all 
of that PSRO's length-of-stay values were at their 
"norm". 



2.5.4 Limitations 

The results presented in this section apply only to 
the twelve PSROs included in the study and not to the 
program as a whole. PSRO selection was based solely 
upon the availability of PHDDS data tapes of accept- 
able quality. To the extent that these data apply to 
older PSROs which may have been more successful 
in reducing unnecessary utilization, they may be ex- 
pected to paint a more optimistic picture concerning 
ALOS changes than might actually be the case. 

A second caveat is that the utilization "changes" 
noted may be due, at least in part, to changes in the 
manner of reporting. A case in point: the data tapes 
for one of the PSROs originally intended for inclusion 
in this study resulted in index numbers so far out of 
line as to indicate gross anomalies in the data. Upon 
investigation, it was found that the more recent tapes 
for that PSRO included absolutely no discharges in- 
volving single diagnoses; this was a clear departure 
from the PSRO's earlier experience as well as from 
medical expectation. This abrupt change in case-mix 
created what appeared to be a marked reduction in 
case-mix adjusted ALOS and an increase in the case- 
mix index. Although it is not known whether this 
anomaly resulted from errors in original discharge re- 
porting or in the computer processing of the data, the 
PSRO in question was dropped from the analysis. The 
extent to which other, less obvious, anomalies may be 
present in the data base or other PSROs is currently 
unkown.'-"* 

A final limitation is that due to time constraints, the 
results presented in this section do not as yet take dif- 
ferences in discharge diagnosis into account. Since 
differences in diagnosis are recognized as the single 
most important determinant of ALOS, the develop- 
mental study will shortly turn to DRGs and/or IGs as 
a further basis for patient classification. The inclusion 
of this additional variable is expected to sharpen the 
analytic potential of the methods described. 



'An important byproduct of this experience is that it 
illustrates another potential use of the index numbers 
described in this section: as a tool, for use by HSQB in 
monitoring the quality of the PHDDS data tapes as 
submitted. 



75 



TABLE 42 
Unadjusted ALOS, 12 PSROs 



Year: 




1977 






1978 




Quarter: 


1st 


2nd 3rd 


4th 


1st 


2nd 3rd 


4th 


Unadjusted ALOS (days) 
Expressed as an index 
(1st quarter 1977=100) 


10.54 
100.0 


10.24 10.08 
97.2 95.6 


10.31 
97.8 


10.61 
100.7 


10.10 10.28 
95.8 97.5 


10.24 
97.1 



2.5.5 Findings 

2.5.5.1 LONGITUDINAL ANALYSIS 

Examination of the composite, unadjusted ALOS for 
the twelve study PSROs combined shows a general 
decline in ALOS between the first quarter of 1977 and 
the fourth quarter of 1978. The relevant figures are 
shown in Table 42. 

The trend may be more seasonal, however, than 
secular. When the data are examined on a quarter-by- 
quarter basis (i.e., first quarter of 1977 vs. first quarter 
of 1978, etc.), the results are mixed: although two of 
the quarters (the second and fourth) show a decline 
in ALOS between 1977 and 1978, the other two do 
not. 

These findings may be modified through applica- 
tion of the case-mix adjustments defined earlier. Ap- 
plication of these techniques indicates that over the 
same period of time, the case-mix index (CM) had in- 
creased in all four quarters while the case-mix ad- 
justed length of stay (CM/ALOS) declined, at a rate 
more rapid than the unadjusted figures would indicate, 
in three of the four quarters. The adjusted and unad- 
justed indices are shown in Table 43 and plotted in 
Figure 14. 

These results suggest that the unadjusted ALOS 
measures associated with these twelve PSROs reflect 
two distinct and separable effects: (a) The case-mix 
proportions of those hospitalized have increased in the 
direction of cases which normally require longer stays. 



TABLE 43 
Unadjusted vs. Adjusted Indices, 12 PSROs 



Unadjusted 
ALOS* 



Adjusted Indices* 



CM 



CM/ALOS 



1st quarter, 
2nd quarter, 
3rd quarter, 
4th quarter, 
1st quarter, 
2nd quarter, 
3rd quarter, 
4th quarter. 



1977 
1977 
1977 
1977 
1978 
1978 
1978 
1978 



100.00 
97.2 
95.6 
97.8 

100.7 
95.8 
97.5 
97.1 



100.0 
102.1 
102.0 
103.1 
102.4 
103.4 
103.5 
103.1 



100.0 
95.1 
93.8 
95.2 
98.2 
92.8 
94.3 
94.1 



while (b) the "true" ALOS for cases of a given type 
has declined more rapidly than might otherwise have 
been thought. The 2.9 percent decrease in unadjusted 
ALOS observed between the first quarter of 1977 and 
the fourth quarter of 1978 was, in other words, the re- 
sult of a 3.1 percent increase in case-mix (CM = 103.1) 
accompanied by a 5.9 percent decrease in case-mix 
adjusted LOS (CM/ALOS =94.1). 

Pursuing these findings at the level of individual 
PSROs, it was found that: 

a. Between 1977 and 1978, the twelve PSROs stud- 
ied showed an increase in case-mix index for 

37 of the 48 possible quarterly comparisons (i.e., 
twelve PSROs times four quarters each), and a 
decrease for only ten of those quarterly com- 
parisons. (One of the quarters resulted in a tie, 
i.e., CM1977 = CM1978.) 

b. Over the same period of time, the twelve PSROs 
showed a decrease in case-mix adjusted length 
of stay for 34 of the 48 possible quarterly com- 
parisons and an increase for only 14. 

These disaggregated findings are highly significant 
from at least two standpoints. First, they are statis- 
tically significant in the sense that they are not likely 
to represent chance phenomena, at least for the 
PSROs studied." Second, they are significant from the 
standpoint of their implilications in terms of PSRO 
review: they denote that the PSROs in question are 
indeed taking on a more demanding mix of cases, i.e., 
a higher percentage of cases which tend to be asso- 
ciated with longer stays; this is in turn a reflection 
that admission review appears to be "working." More- 
over, when the ALOS is adjusted to take this change 
into account, sharp declines in adjusted length-of- 
stay are noted, indicating that continued stay review 
has also had an effect.^^ 

When case-mix adjustments were applied to Medi- 
care and Medicaid discharges separately, patterns 



1st quarter 1977=100 



"'The test of significance employed was a binomial sign test 
(N=48). In botfi cases, i.e., CM and CM/ALOS, the level of 
significance was less than .005. 

"- In the absence of a suitable comoarison group, there is no 
way of knowing whether these effects are unique to the 
PSRO areas studied or whether they might not also apply 
to all areas, with or without PSROs. The purpose Df this 
analysis, however, is not to study the PSRO effect per se 
but rather to indicate how that effect might be more clearly 
illuminated. 



76 



Figure 14 

Unadjusted vs. Adjusted Indices, 12 PSROs 
(All Patients Combined) 




u u 

(U CO 

U 01 

U >-i 

3 

c 



77 



similar to the foregoing were observed; that is, for 
both Medicare and Medicaid a general increase in 
case-mix index was accompanied by a decrease in 
case-mix adjusted length of stay. The effect was much 
more pronounced, however, for Medicare than for 
Medicaid. Medicare and Medicaid findings comparable 
to the aggregate figures presented earlier are shown in 
Table 44 and plotted in Figure 15 and 16. 

When disaggregated to the PSRO level, the following 
additional findings were noted: 

a. Medicare — Of the 48 possible quarterly com- 
parisons involving the case-mix index, 34 showed 
an increase between 1977 and 1978. Of the 48 
comparisons involving case-mix adjusted length 
of stay, 32 showed a decrease. The former find- 
ing is significant at the .005 level of significance; 
the latter is significant at the .02 level. 

b. Medicaid — of the 48 possible quarterly compari- 
sons involving the case-mix index, only 25 
showed an increase. Of the 48 comparisons in- 
volving case-mix adjusted length of stay, only 



TABLE 44 

Unadjusted vs. Adjusted Indices, 12 PSROs 
by Beneficiary Group 



Beneficiary 


Unadjusted 


Adjusted Indices* 


Group/Quarter 


ALOS* 


CM 


CM/ALOS 


Medicare: 








1st quarter, 1977 


100.0 


100.0 


100.0 


2nd quarter, 1977 


96.9 


101.6 


95.6 


3rd quarter, 1977 


95.2 


101.6 


93.7 


4th quarter, 1977 


96.9 


102.4 


95.0 


1st quarter, 1978 


101.2 


102.5 


98.5 


2nd quarter, 1978 


95.2 


102.8 


92.7 


3rd quarter, 1978 


97.8 


103.7 


94.5 


4th quarter, 1978 


96.3 


102.5 


93.8 


Medicaid: 








1st quarter, 1977 


100.0 


100.0 


100.0 


2nd quarter, 1977 


97.4 


101.9 


95.8 


3rd quarter, 1977 


97.5 


101.7 


96.2 


4th quarter, 1977 


99.2 


101.2 


97.9 


1st quarter, 1978 


99.9 


100.7 


99.4 


2nd quarter, 1978 


97.0 


101.1 


96.3 


3rd quarter, 1978 


96.8 


100.6 


96.1 


4th quarter, 1978 


97.2 


100.2 


97.1 



25 showed a decrease. Neither of these findings 
is significant. 

2.5.5.2 CROSS-SECTIONAL COMPARISONS 

Although of lesser relevance than the longitudinal 
findings presented above, comparisons across PSROs 
are of value in arriving at an understanding of the 
ALOS differences among PSROs and in observing how 
much of that variation is attributable to differences in 
case-mix and how much to other factors. 

In the last quarter of 1978, for example, the unad- 
justed lengths of stay associated with the twelve 
study PSROs ranged from a low of 6.8 days (in Cali- 
fornia) to a high of 12.6 (in New Jersey). The PSRO 
with the highest ALOS, in other words, had an unad- 
justed ALOS 85 percent higher than the PSRO with 
the lowest. 

Case-mix adjustment disclosed, however, that much 
of that difference stemmed from differences in case- 
mix; specifically, the case-mix index (CM) of the New 
Jersey PSRO was estimated to be 1 14.4 as compared 
to 87.0 for the California PSRO, a difference of 31 
percent. Because of this difference in case-mix, the 
adjusted (CM/ALOS) indices for the two PSROs dif- 
fered by only 44 percent (106.5 vs. 74.2). The ap- 
parent 85 percent difference in ALOS, in other words, 
can be disaggregated into two components: a 31 per- 
cent difference due to case-mix and a 44 percent dif- 
ference due to true differences in length of stay.»=* 
Similar effects were noted for other PSRO pairs and 
other periods of time. 

2.5.6 Conclusions 

The methods suggested in this section would ap- 
pear to offer a suitable basis for providing an under- 
standing of ALOS performance not possible through 
the use of aggregate data alone. In particular, these 
methods permit ALOS changes to be factored into two 
separate components: one component attributable to 
changes in case mix, the other to differences in true 
ALOS behavior. As noted, the process can be applied 
either longitudinally to study trends over time, or cross- 
sectionally to study differences among PSROs or 
groups of PSROs. 

Extensions of these methods to incorporate diag- 
nosis-specific information will be applied in further 
work and are expected to yield additional useful 
insight. 



1st quarter 1977=100 



Ignoring minor errors due to rounding, a 31 percent 
difference compounded on top of a 44 percent difference 
produces an 85 percent difference overall, i.e., 1.44 times 
1.31 equals 1.85 (approximately). 



78 



Figure 15 

Unadjusted vs. Adjusted Indices, 12 PSROs 
(Medicare) 




^1 !-i 

01 as 

j-i 0) 

U >-i 

D 

o- 



79 



Figure 16 

Unadjusted vs. Adjusted Indices, 12 PSROs 
(Medicaid) 




80 



3. Benefit-Cost Analysis of Concurrent 
Review 

3.1 Objectives 

This section presents a benefit-cost analysis of 
PSRO concurrent review as it pertains to Medicare 
beneficiaries. The estimate of "days saved" due to 
PSRO review generated by the Medicare Impact Study 
(Section 2.2) is used as input into calculations which 
provide an estimate of reductions in Medicare pro- 
gram reimbursements. Reductions in Medicare pro- 
gram reimbursements are compared to estimated cost 
of PSRO concurrent review. The study's time frame is 
CY 1978. As was mentioned in Section 1.8, it may not 
be appropriate to extend CY 1978 results to a fully 
implemented program. 

The benefit-costs presented in this section relate in- 
cremental benefits to full costs. The Medicare Impact 
Study estimates the benefits of PSRO review relative 
to utilization review (UR). Comparable incremental 
costs could not be calculated because reliable esti- 
mates of UR costs do not exist. If UR benefits are 
small, the approach is essentially correct since it very 
closely approximates a matching of full benefits to full 
costs. To the extent that UR benefits or UR costs are 
substantially greater than zero, the results of the 
analysis are biased. 

Given the significant coeffificient of the PSRO-re- 
gional interaction variable in the Medicare Impact 
Study, separate benefit-cost estimates are calculated 
for each region of the country. Regional effects are 
aggregated to provide a national estimate. National 
benefit-cost estimates are the primary result of this 
section. 

The remainder of this section presents the develop- 
ment of Medicare concurrent review costs and bene- 
fits, an appraisal of factors excluded from the analysis, 
data sources, and findings. 



3.2 Medicare Concurrent Review Costs 

The denominators in the benefit-cost analyses are 
concurrent review costs for Medicare patients in 
short-stay hospitals during CY 1978. These costs 
include an apportioned part of PSRO "overhead" 
costs. A cost figure is calculated for each Census 
region and for the nation. The data for calculating 
cost components come from an updated Unit Cost 
Report for CY 1978 and yearly counts of Medicare 
and total discharges by PSRO, both of which are pro- 
vided by HSQB.i 

Equations (1) & (2) demonstrate the calculation of 
Medicare review costs: 



(1)TMRC = 
(2) ALLOC 



%MED * [TCR Cost + ALLOC] 

TCR Cost 

* [AREAC + MSC] 



TCR Cost + MCEC 



where TMRC = Total Medicare review cost in dol- 
lars by PSRO 
%MED = % Medicare discharges 
TCR Cost =^ Total concurrent review costs 
ALLOC = Allocated portion of areawide and 

management/support costs 
MCEC = Total MCE review costs 
AREAC = Areawide review costs 
MSC = Management/support costs 

The Medicare review costs are calculated for each 
PSRO and then summed to give the total Medicare 
review cost for each of the four Census regions and 
the nation. 

The following concurrent review costs were com- 
puted for Medicare patients for each Census region 
and the nation: 



^The Unit Cost Report is described in Section 4.2. 







TABLE 45 
Medicare Review Costs in 1978 






Region 


Total 

Medicare 

Concurrent 

Review Costs 


No. of 
Enrollees 




Cost per 
Medicare 
Enrollee 


No. Medicare 
Discharges 


Cost per 
Medicare 
Discharge 


Northeast 
North Central 
South 
West 


$23,421 ,471 
16,987,578 
18,787,833 
18,473,903 


4,728,898 
3,783,674 
3,383,325 
2,958,974 




$4.95 
4.49 
5.55 
6.24 


1,525,058 

1 ,434,769 

1,200,597 

977,931 


$15.36 
11.84 
15.65 
18.89 


Nation 


$77,670,785 


14,854,871 




$5.23 


5,138,355 


$15.12 



81 



The total Medicare current review costs in Table 45 
are the denominators for the benefit-cost ratios. To 
provide additional descriptive comparisons among the 
regions, the cost per enrollee and cost per discharge 
are also shown in Table 45. Review cost per enrollee 
varies from $4.49 in the North Central to $6.24 in the 
West while cost per discharge ranges from $11.84 to 
$18.89 in the same two regions. 



3.3 Medicare Concurrent Review Benefits 

This section develops the Medicare concurrent re- 
view benefits required for the benefit-cost analysis. 
The calculation of benefits is first shown to be based 
on the Medicare reimbursement formula. Substitution 
effects which offset Medicare program benefit esti- 
mates are then presented. The section concludes with 
a discussion of benefits and factors excluded from 
the analysis and a discussion of how Medicare re- 
imbursements relate to broader societal concerns. 

3.3.1 Calculation of Benefiits 

The Medicare program benefits estimated for this 
section are reductions in Medicare reimbursements 
which are equal to the number of Medicare days saved 
multiplied by "the value of a day saved." This section 
utilizes the Medicare reimbursement formula to cal- 
culate the gross value of days saved. The gross value 
of a day saved is partially offset by the cost of in- 
creased utilization of long-term care and ambulatory 
care services. These offsets are estimated by the cal- 
culation of substitution effects which reduce the esti- 
mated gross value of a day saved. 

3.3.1.1 FOUR PRIMARY SOURCES OF BENEFITS 

Gross Medicare reimbursement savings associated 
with the value of day saved estimates are comprised of 
four effects: (1) routine reallocation; (2) routine vari- 
able cost; (3) ancillary reallocation; (4) ancillary vari- 
able cost. These effects are derived directly from the 
Medicare reimbursement formula: 

Ri = Mi/(Mi + Ni) * TREj + MCi/(MCi + NC;) * TAEj 

where Rj = annual reimbursements to hospital i 
Mj = annual Medicare days of care in hos- 
pital i 
Ni = annual non-Medicare days of care in 
hospital i 
TRE, =annual routine services expenditures for 

hospital i 
MCi^ annual Medicare ancillary charges in 

hospital i 
NC, = annual non-Medicare ancillary charges 
in hospital i 
TAEi = annual ancillary services expenditures 
in hospital i 



The derivation of each effect is presented below. 

Routine Reallocative Effect: A decrease in Medi- 
care utilization will cause a reduction in reimburse- 
ment in the short run because the reduction in the 
ratio of Medicare to total days [Mi/(Mi + N,)] 
changes the apportionment of total routine expendi- 
tures, reducing the Medicare share of TRE. 

Routine Variable Cost Effect: With decreases in 
Medicare utilization, the total routine expenditures 
may be reduced by the amount of variable costs ac- 
counted for in the average per diem for routine 
services. As the variable costs decline, so does TRE, 
and these reductions in expenditures reduce Medi- 
care reimbursements by an apportioned amount 
[Mi/(Mi + Ni) * A TRE]. A key assumption in the esti- 
mation of the routine variable cost effect is the ap- 
portionment of the average per diem into variable and 
fixed costs according to the 40:60 ratio which is gen- 
erally accepted in the hospital cost literature.^ This 
variable/fixed cost ratio is assumed to be constant 
irrespective of the level of total days of care or the 
percentage of Medicare days in the hospital. Since 
this is a short-run analysis of reimbursement savings, 
input factor prices and production technologies are 
assumed to be unaffected by the reduction in Medi- 
care days. 

Ancillary Reallocative Effects: The reimbursement 
savings resulting from the reallocation of ancillary 
service expenditures operates in an analogous way to 
the routine reallocation effect. Reductions in Medicare 
days reduce the ratio of Medicare ancillary charges 
to total ancillary charges which reduces the amount 
of ancillary expenditures allocated to the Medicare 
program. 

Ancillary Variable Cost Effect: Just as the saving in 
Medicare days reduces total routine expenditures, it 
also reduces the ancillary service expenditures by the 
amount of variable costs associated with the ancillary 
costs per diem multiplied by the number of days saved. 
The 40:60 assumption on the variable/fixed cost ratio 
applies here as well. It is necessary to assume that 
each day saved has the same volume of ancillary 
charges per day as the average for all Medicare days 
in the total stay. In a way similar to the routine vari- 
able cost effect, the reduction in ancillary variable 
costs reduces total ancillary expenditures, and these 
savings are apportioned to the Medicare program by 
the ratio of Medicare to total ancillary charges. 

Given the percentage reduction in Medicare days 
for each Census region (derived from the Medicare 
Impact Study), the cfianges in reimbursements are 
calculated for each of the four components described 
above. This is accomplished by essentially substituting 



- For example, see Joseph Lipsomb, I. E. Raskin, and Joseph 
Eichenholz, 'The Use of Marginal Cost Estimates in Hospital 
Cost-Containment Policy," in Michael Zubkoff, I. E. Raskin, 
and Ruth S. Hanft, Editors, Hospital Cost Containmer)t: 
Selected Notes for Future Policy (New York: Prodist, 
1978). 



82 



hospital data into tlie l\/iedicare reimbursement formula. 
Each cost component is summed across hospitals to 
get the totals for the four effects at the region level, 
and the sum of the four components are the total 
gross savings before adjustment for substitution 
effects. 

3.3.1.2 SUBSTITUTION EFFECTS: OFFSETTING 
GROSS BENEFITS 

Two substitution effects, one for long-term care 
(LTC) services and another for ancillary services, 
partially offset the savings from the four primary 
effects outlined above. 

LTC Substitution Effect: It is assumed that 7 percent 
of short-stay hospital stays are followed by a stay in an 
LTC facility. For purposes of calculating the LTC sub- 
stitution effect, it is assumed that for every 100 days 
saved in the hospital. Medicare must pay for an addi- 
tional 15 days of LTC. The reduction in savings due 
to the long-term care substitution effect is calculated 
by multiplying this substitution factor (0.15) by the 
number of days saved in each region and by the aver- 
age LTC per diem in the region. 

Ambulatory Substitution Effect: Reductions in days 
of care will lead to increases in ambulatory care be- 
cause it is assumed that outpatient ambulatory serv- 
ices will be substituted for inpatient ancillary services. 
The majority of ancillary services occur early in a 
hospital stay. Since it is impossible to empirically de- 
termine whether the l\/ledicare days saved are from the 
beginning, middle or end of the stay, no empirical 
basis exists for establishing the ambulatory substitu- 
tion rate. Therefore, it is assumed that the ancillary 
charges associated with each day saved are equal to 
the mean ancillary cost of all Medicare days. In cal- 
culating the ambulatory care substitution effect, one 
must decide what proportion of hospital ancillary 
services will be replaced by outpatient services. A 
0.5 rate was used last year based on recommenda- 
tions of health practitioners and researchers. This 
assumption is controversial because it has a strong 
influence on the benefit estimates. This year it was 
decided to produce three separate sets of benefit 
estimates based on ambulatory substitution ratios of 
0.3, 0.5, and 0.7. The substitution effect is calculated 
by multiplying the inpatient ancillary reimbursements 
per diem by the number of Medicare days saved, and 
by the substitution rate factor. 

3.3.1.3 POTENTIAL BENEFITS AND FACTORS 
EXCLUDED FROM THE ANALYSIS 

This study is limited to measures of the benefits of 
PSRO review to the Medicare Program. Benefit-cost 
estimates are not made for Medicaid and Title V bene- 
ficiaries because data were not available. The lack of 
appropriate data also accounts for the fact that PSRO- 
induced savings in Part B reimbursements are not in- 
cluded in the benefit calculations. 



The potential for savings in Part B will vary consid- 
erably depending on whether the PSRO savings result 
from reduced length of stay or reductions in admis- 
sions. If the savings occur as reduced length of stay, 
it is unlikely that there will be much change in the 
number or type of physician services or procedures 
performed. However, if the utilization reduction occurs 
as reduced admissions, one would expect the amount 
of physician care and reimbursements under Part B to 
be reduced significantly and to vary with the type of 
admission eliminated. 

Other factors excluded from the benefit-cost anal- 
ysis are the benefits of areawide and MCE review. 
The benefits of health care equality assessment, and 
the benefits of physician and patient profiles are not 
considered in the analysis. An exploratory benefit- 
cost analysis for MCE activity is presented later in this 
report (Section 5.5). However, the results of that 
analysis are too preliminary to be included in a na- 
tional PSRO benefit-cost analysis. 

In addition to the four primary effects and two sub- 
stitution effects described above that affect Medicare 
reimbursements, variations in hospital occupancy 
rates, length of stay, and cost per discharge may 
modify the PSRO effects upon Medicare reimburse- 
ments. However, last year's simulation analysis indi- 
cated that these three hospital characteristics have 
minor and unsystematic effects on the level of reim- 
bursements. Hence, these factors are not taken into 
account in calculating the value of a day saved. 

3.3.2 Social Benefits Associated with PSRO Medicare 
Concurrent Review 

3.3.2.1 TRANSFER OF COST BURDEN FROM 
MEDICARE JO NON-MEDICARE 
POPULATIONS 

One potential impact of the PSRO program on the 
non-Medicare population is that the application of the 
hospital reimbursement formula may transfer financial 
burden from Medicare program beneficiaries to non- 
Medicare patients. Because part of the Medicare reim- 
bursement savings derived from the allocation effects 
are in fact transfers of fixed cost to non-Medicare pa- 
tients, reductions in Medicare days will increase the 
costs to non-Medicare patients. It should be stressed 
that the analysis reflects short-term behavior. If hos- 
pitals respond to decreased utilization by reducing 
their fixed costs, then the long-run benefits to the 
Medicare program will increase and the transfer of 
fixed costs to non-Medicare patients will be reduced. 

3.3.2.2 PSRO IMPACT ON HOSPITAL UTILIZATION 
OF NON-FEDERAL PATIENTS 

The Congressional Budget Office takes issue with 
the estimated PSRO impact calculated in last year's 
evaluation. 3 One of the differences in estimates con- 



' Congressional Budget Office, "The Effect of PSROs on 
Healtli Care Costs: Current Findings and Future 
Evaluations." 



83 



cerns the impact of the "Roemer" effect. This effect 
(named after l\/lilton Roemer, who first suggested the 
concept) is based on the hypothesis that increases in 
short-stay beds will lead to increases in short stay 
utilization. Basically, the hypothesis suggests that sup- 
ply creates its own demand. There is considerable 
evidence in the economic literature to suggest that this 
effect does exist. The Medicare Impact Study also 
found a positive correlation between bed supply and 
utilization rates. CBO concludes from this theory and 
research that reductions in Medicare utilization free 
up beds and that this is equivalent to increasing the 
bed supply. Therefore, they conclude that decreases in 
Medicare use will be offset by a concomitant increase 
in use by the private sector patients (their estimate is 
40%). The Roemer effect would also cause a sec- 
ondary impact on Medicare reimbursements through 
the reallocation effects for the reimbursement formula. 
(Reimbursements would decrease as the percent 
Medicare declines.) 

However, it is not apparent that the transfer of the 
effect of bed supply upon utilization to the PSRO case 
is a direct one. It can also be argued that there are 
spill-over effects of PSRO review. This argument im- 
plies that physician and hospital behavior are not 
sharply differentiated between Federal and private 
sector patients. That is, if physician behavior is modi- 
fied due to PSRO review with regard to Federal pa- 
tients, it is likely that their behavior toward private 
sector patients will also change. If the Roemer effect is 
operative, the PSRO effect will be dampened. If the 
spill-over effect is operative, the PSRO effect will be 
enhanced. If both are operative, they will negate each 
other to some extent. 

To date, there is no direct evidence for either effect. 
The Office of Research, Demonstrations and Statistics 
is presently conducting an analysis to test for Roemer 
and/or spill-over effects. The analysis is essentially a 
replication of the 100% Medicare Claims File study, 
using data on all patients, both Federal and private. 
Until this analysis is complete, there is little reason 
to include either Roemer effect or spillover effect in 
the analysis of PSRO benefit costs. 



3.4 Data 

The data on the days-of-care saved per 1,000 eligi- 
bles for the four Census regions were obtained from 
the Medicare Impact Study. The principle source of 
data for the calculations of the value of a day saved 
is the file of audited provider cost reports for FY 
1978 and FY 1979. This file contains information on 
days-of-care and expenditures for routine and an- 
cillary services far both total admissions and Medi- 
care admissions. The long-term care per diem reim- 
bursements required for one of the substitution effects 
were derived from State data contained in a HCFA 
Note on the Utilization of Hospital Services in 1977. 



3.5 Findings 

The findings of the benefit calculations are presented 
in three parts. First, the results are discussed for the 
four primary sources of savings arising from the 
reimijursement formula. Second, the calculations for 
the long-term care and ambulatory substitution effects 
are discussed; and finally, the potential transfer of 
fixed costs from the Medicare program to non- Medi- 
care patients is described. 

3.5.1 Reimbursement Savings for Four Primary Effects 

In the 108 active PSRO areas included in the bene- 
fit study, the average Medicare per diem short-stay 
hospital reimbursement in FY 1978 was $188.79, ap- 
proximately $19 more than reported in last year's 
evaluation. The total per diem reimbursements (includ- 
ing ancillary services) range from $155.50 in the South 
to $243.42 in the West. Ancillary reimbursements per 
diem were $89.94 for the nation as a whole, $76.46 in 
the South, $86.01 in the Northeast, $89.01 in the North 
Central, and $122.72 in the West. These ancillary re- 
imbursements account for from 44 to 50 percent of the 
total per diem in the four regions. This is a consider- 
able increase from the average ancillary per diem of 
$62.37 reported last year (36.7% of total per diem). In 
the following tables many of the results are reported as 
the savings per day expressed as a percent of the 
total per diem. 

Table 46 presents the gross savings derived from 
the Medicare reimbursement formula components (i.e., 
before accounting for substitution effects). Since the 
PSRO activity in the South resulted in a net addition 
to days of care, the table shows negative savings 
components for this region. The reallocation effect 
savings for routine services is $64.81 for the nation 
or 34.3 percent of the total per diem. This corresponds 
to the finding of 37.1 percent in the 1978 evaluation. 
For the four Census regions, the routine reallocation 
effects range from 28.8 to 31.9 percent of the per diem. 
The variable cost effect was found to be 1 1.31 per- 
cent of per diem or $21.36 for the nation. In all, the 
reduction in reimbursements for routine services 
amounts to 45.6 percent of th© total per diem for the 
nation. 

The reallocation effect for ancillary services 
amounted to $60.86 savings or 32.24 percent of per 
diem for the nation. This is significantly higher than 
the $38.14 and 22.5 percent reported last year. The 
large increase in the share of ancillary services of the 
total per diem has made ancillary services a more 
significant source of Medicare savings. The variable 
cost effect for ancillary services was $20.09 per day for 
the nation or 10.64 percent of total per diem. The total 
reimbursement savings associated with the ancillary 
effects was $80.95 or 42.88 percent of the total per 
diem. 

The total gross savings due to the components of 
the Medicare reimbursement formula is the sum of the 
results for the four effects described above. For the 



84 






J2 

3 

E 

o 
u. 

3 
.Q 

E 
'S> 



■o 
a> 



CO *- 



3 
X) 



M 



> 
(O 



o 






o CO I- E 
< 



O 3 v^ 

(0 m c 

^ © U- 



tn 
O 
O 

n © 

(0 ^ 
■c UJ 

CO 

> 



c 
o 

CO ^ 

_0 (1) 

CO LU 
CD "^ 
QC 

CO 



o 

o 

^ CD 
CO ^ 
C lU 

CO 

> 

C\j 



7J 


c 


O 


o 


cc 






O 








Slu 



CO 

c 



E 

CD 

CO Q 



c _« 



o _ 

^o CO 1- E 

< 



c - E 
CO Q 



(ot-°-Q 
< 



to 

c ir E 

CD Q 

CO 



o _ 

< 



^^ 



en 

.E 

co 
CO 



<0 1— LJ 



(0 

CO Q 
CO 



w c 

^ o 
« — 

C O) 


ODC 



-^ -^t 1- CO 


lO 


00 1- T- CO 


00 


r- 00 00 1^ 


00 



00 CD -^ 1- 
C3) O T- 00 

CD 05 CC) O 

in ■<a- CM CJ5 



Tt O 00 1- 
00 CM -I- O 
00_ T-_ tT CD 

r-" co" cd" o" 

CM CO CM C\J 

o> C3) in T-_ 
■<*■" oo" r^" cm" 

Tj- •«3- in CM 



</J 



vP vP vP vP 

p^ p^ tf^ o^ 
CD •* 00 1^ 
CO CD ■>* T-_ 

ai ai ao ai 



T- ■.- C3) CO 

O 00 •!-_ CO 

00 in CO CM 

1- 1- 1- CM 

<«■ I 



T- CM •^ 1- 

•I- Ti- T-; CO 

OCJ CM Cvi CM 
CM CO CO CO 



CD ■>+ h- CD 
O CO C3) CD 
tT oi C3> CO 

in in ■«3- h~ 



in CO CO CO 

in <3) CM t-; 

cr> in CO o> 



CD O -1- •■- ^ 

T- r«- in T- CO 

<I) OC) OCJ CC3 T^ 



CO ^- C35 en 
00 CD C3i r- 


CO 
CO 


O T- T- CO 

CO CO CO CM 


CO 


CD 00 in 00 
CO en r- o 


00 


en r^ cj) o 
in in ■<«• r^ 
w 1 


s 



mo -, 



85 



o 

Ul 

c 
o 



(A 

JO 
3 
V) 

»s 

•«t 5 

(0 

E 

< 

c 

(0 



(0 

O 

E 

bO 

c 
o 



00 

< 



m 

c 



LU 



=3 

w 
o 

£ 
< 





H- 
I 

O) 

c 
o 



vO « E 

o^ o 9J 
«•- Q 

*^ o fe 



vP vP nJ5 vP vP nP vP sP vP vP vP vp 

T-COOOi-'^LnoOCvjCjT-Oi 
■>^C0COi-;O>(DI~~;LOTl;C\jC0C0 
OOJoOi—OOOOOt-'t-^t-^ 



O 

Q 



« E 
I-Q 
O 





5 4 



E 




CO 



O 
Q. 



O 
O 



CO t: 

H- 1- 
O 

Q. 



So 



\- Q 

'o 
a. 



o 

Q 



CO 

. CO 
O (/} 



w c 

C O) 


OCT 



I + + + 



00 1- t in CD 

in N- 00 in o 

o co_ c\j o_ o 

o" cj in oo" oo" 

•^ in CD CD T- 

CD T- CD 1^ CD 

K ■*" o" cvT co" 

CO m r^ 1- 1- 



CD CM O en C\J CM 1- 

in CD 1- in 00 ■^ o 

°l CO -I- 00^ 00_ CO 00_ 

r-" T-" in oo' cjf in o" 

CD -I- 00 in r^ CM h- 

■«* h- CD CD C35 00 CO_ 

Tt ■«;]•" T-" oo" in oo" T-" 

CM T- CM CM 



I I I I I I + + + 



vPnPsPvPvPnP^nPvP^vP^ 
o^ o^ o^ o^ o^ o^ o^ o^ o^ o^ o^ o^ 
T-CDOCJl-i-'^inoOCMCMi-CJ) 

'^cocoincoor^in'«rT-;CMCM 
cocMt-^Tf'^fTf'^'jf-^iniriiri 

•i-CMCOt-CMCO-i-CMCO-i-CMCO 



00 T- 

1- CO 

Oi in 

00"t-" 
CD CO 

r^ CM 

•*"t-" 
CM ^ 



^f CD 
Tl- CM 
1— <J> 

■* 5"" 
c:^ r^ 

^- h-_ 
h-"oo" 

in I 



r~-t-~-c\iocJ)0)CJ)oo 

t^CMCMh-T-OO^rO 
00 00_ CD_ CD T- T-_ CD^ T-_ 

cm" -^" O" -"d^ oo" oo" CO" C3f 
TT^-CDCOOCDT-in 

CO -^^ ■*_ 'd- •<a- CM_ T-_ ai 
rf o" o" h-" Tt" •>*" h~" ai 

I- CM T- I- CM 



+++ 



?^ 


CM 
CD 


■^ 
■^ 


00 



</> 



LU 



OO 



CD 
CD 



CJ) 
CM 

CO 

ay 

CJ) 

co" 



in 

CM 

-ti- 



cs 



CO 

oo" 

CM 
CO 



O 



a> 



CD 
CD 



0"~ 

o 

CD 



O 

o" 

in 

CM 

■*" 

+ 



CD 
CJ> 

CJJ 
CO 



CvJ 
CO 



Oi 

CO 

o 

co" 
m 



W 



d 

m 



CM 

cm' 

CM 



CD 



CO 
CJ) 
CD 






CO 

••a- 
oci 



CO 

CO 

C3) 

in" 



vv vV v» 

o~» O^ O^ 

00 CD •>* 
CM ■<«; CO 
CO CO CO 
CM CO Tf 



CO O CM 
CO O 00 
CO CJ) 1- 

co" o" in" 

r^ T- -"a- 

CD CJ) 1- 

T-" cd" oo" 
Tf in r^ 



vp vO vP 
O^ O^ tf^ 

00 CD ■>* 
CM ^ CO 

iri iri iri 

1- CM CO 



t^ r^ o 

1- 00 CO 
■«a- CD CJ) 

T-" in" CJ) 
in 00 T- 
co in oo_ 
h~" in co" 

CM TJ- CD 



T3 

0) 
Q. 



c 

0) (/) 

E c 

<D (D 

52 E 

3 Q) 

£i CO 

E 3 

If 

a> 0) 



•O CO 
0) 

s ^ 

Q. c 
v> — 

^§ 

^^ 

o "" 

SI <D 

tn x: 

cc o 
o "^ 



o £; 
c ° 

' 0) c 
O CO 

0) <n 
a. CO 



. J3 
E E 

■o "- 

>_ 0) 



<D 

E 

O) 
{0 

3 
Si 

E 

S 

^ 

E 
.2 

(D 
Q. 



a 
tn 
o 



CO 

o 

o 



S.<o;:i ^ 



CO "O - 
*- 0) o 

CO 

O C CO 

CO O „_ 

2Z° 



So 
>,o 

CO 



c 

0) 

o 

Q. 

CO 

CO 



"O CO 

^1 



TO « 

CO I. d) t 

•O CD O. CU 

•- °-C0 C 



o >> 

CO O Q. 

CO — ns 

i2 = 

ci) o "D 

CO <»- 

c B 

o o 



^- yi* ■•-• C en 



1 .9 



.i 5 

^^ 

-■ CO 

1^ o 
CO "O 
o _ 
•O CO 



J2S^^ 
o S 52^ 

TJ •- 3 = 

*- E •- £" 
o~ £ o 

o = — fc 

— O CO 

c 

CO 



"D 5 "C 

C (0 ^ 

0) O CO 
c «•§ 

O 3 CO 



■•5 O 

3 C 

•:5 CO ~ J 

^ £ £o 

3 CO CO "a 

« o Ci 



22k J 



Q XJ TJ CO 



lio 

(O CO ^ 
3 O (0 

^ EO 

E 3 I- 



86 



nation as a whole, the gross savings amounts to 
$167.12 or 88.52 percent of the total per diem. These 
gross savings are subject to reductions from the sub- 
stitution effects discussed below. 

3.5.2 Long-Term Care and Ambulatory Substitution 
Effects 

In Table 47 the magnitude of the long term care and 
ambulatory substitutions effects are described for the 
nation and for each of four Census regions. The figures 
in the "total offset" column show the amounts by 
which the total gross savings of Table 46 must be re- 
duced to give the net savings adjusted for the sub- 
stitution effects. In the case of the South, the offsets 
are positive because they in effect increase savings 
by reducing the negative savings reported in Table 46- 

The reductions for the long-term care effect are 
calculated assuming a substitution rate of .15 nursing 
home days for every 100 hospital days. For the nation, 
the long term care per diem reimbursements are $80.00 
which is only 42.38 percent of the short-stay hospital 
per diem. The total national offset to savings for the 
long term care substitution effect is $14,325,222 which 
translates to 8.0 percent of the total per diem. 

The ambulatory substitution effect is calculated for 
three substitution rates: 0.3, 0.5, 0.7. The reduction in 
reimbursement savings due to the ambulatory substi- 
tution effect amounts to $27.3, 45.6, or 63.8 million for 
the nation depending on the rate of substitution. This 
means the ambulatory substitution effect can range 
from 15 to 35 percent of the total per diem. The sum of 
both substitution effects, therefore, will cause the po- 
tential savings in reimbursements to be reduced by 
$41.7, 59.9, or 78.1 million which is 23.3, 33.4, and 43.6 
percent of the total per diem. 

Since there is a Part B savings component omitted 
from the calculations of inpatient reimbursement sav- 
ings, the net increase in Part B reimbursements due to 
the increase in outpatient or ambulatory services will 
be less than the amount suggested by the arnbulatory 
substitution ratio. Hence, one way of compensating for 
the unmeasured inpatient Part B savings would be to 
calculate the dollar offset for the ambulatory substitu- 
tion effect using a 'slightly lower' substitution ratio. 
Consequently the net increase in reimbursement for 
ambulatory services may be closer to $41.7 million 
than $59.9 or $78.1 million. 

3.5.3 Potential Shift of Fixed Costs to Non-Medicare 
Patients 

The calculations in Table 46 based on the reimburse- 
ment formula show that for each day saved, the reallo- 
cation effects for ancillary and routine services re- 
duced Medicare reimbursements to hospitals by 66.57 
percent of the per diem. The variable cost effects indi- 
cate a reduction in Medicare reimbursements equal to 
21.95 percent for each day saved for a total of 88.52 
percent for all four effects. It is evident that the opera- 
tion of the reimbursement formula apportions changes 
in costs between the Medicare program and non- 



Medicare sources of payments. If, as is widely be- 
lieved, the variable costs are equal to 40 percent of 
per diem, and if total inpatient reimbursement savings 
are 88.52 percent of per diem, then up to 48.52 percent 
of the per diem savings to Medicare must be due to 
a transfer of the burden of fixed costs to non-Medicare 
patients, the hospitals, or third parties. 

Table 48 illustrates the changes in savings or ex- 
penditures at the margin for a one-day reduction in 
Medicare days at each of the four regions and the 
national levels. If one assumes that the full amount of 
the change in variable costs is realized as savings to 
either Medicare or non-Medicare patients, then sub- 
tracting the variable costs savings for non-Medicare 
patients (column 5) from the total reallocation effects 
(column 6) will give the maximum burden of costs that 
may be transferred from Medicare reimbursements to 
the non-Medicare sector (column 7). Hence, if one 
Medicare day is saved in the Northeast region, up to 
$74.04 of the reductions in reimbursements may in 
fact be transferred fixed costs to non-Medicare pa- 
tients causing an increase in their expenditures per 
day (i.e., the price of hospital care goes up for non- 
Medicare patients). In the West, the maximum amount 
of transferred fixed cost is $93.43, and for the nation it 
would be $91.60 or 48.52 percent of the per diem. 

The transferred fixed cost of $91.60 multiplied by 
the number of Medicare days saved (948,430) gives 
$86.9 million, the maximum fixed costs transferred 
from the Medicare program to non-Medicare patients 
for the nation. Since there were 175.8 million non- 
Medicare hospital days during 1978, the increase in 
the average price per day for non-Medicare patients 
is $0.49 which is an increase in the average per diem 
of 0.2 percent. 

3.5.4 Benefit-Cost Analysis Results 

The purpose of this section is to combine the re- 
sults of the Medicare review cost analysis and the 
computation of benefits for Medicare review to gen- 
erate the appropriate benefit-cost ratios. Separate 
benefit-cost ratios are calculated for each of the four 
Census regions and for the entire nation. The total 
concurrent review costs for Medicare admissions in 
1978 (Table 45) are the denominators, and the bene- 
fits, the numerators, are the total savings from the 
four primary effects shown in Table 46 reduced by the 
total offsets for the two substitution effects described 
in Table 47. 

Table 49 displays the derived benefit-cost ratios. A 
separate set of ratios are produced for each of the 
three assumed ambulatory substitution rates (0.3, 0.5, 
and 0.7). As mentioned in a previous section, the bene- 
fits are the incremental benefits for PSRO review be- 
yond the savings produced through utilization review. 
However, the costs are the full costs of PSRO review 
for Medicare patients. Since the PSRO review costs are 
not reduced by the amount of utilization review costs, 
the benefit-cost ratios may be biased downward. In 
other words, the ratios of Table 49 represent con- 



87 



servative estimates of the "true" benefit-cost ratios. 
The existence of unmeasured Part B savings may be 
invoked as a justification for favoring the benefit-cost 
ratios based on one of the lower ambulatory substitu- 
tion rates (either 0.3 or 0.5). 

If the Roemer effect is important, there may be in- 
creases in non-Medicare utilization that will increase 
Medicare reimbursements through the variable cost 
effects, and decrease reimbursements through the re- 
allocation effects. Finally, these benefit-cost ratios do 
not take into account the fact that part of the savings 
to the Medicare program may result in increased ex- 
penditures to other patients. 

The benefit-cost ratios in the 1978 evaluation ranged 
from 0.55 to 1.65 under alternative specifications, but 
since the results clustered in the neighborhood of 1.1, 



that was the finding reported. As shown in Table 49, 
the substitution rate used last year (0.5) yields a na- 
tional benefit-cost ratio of 1.269 this year. That is, each 
$1 spent in PSRO review led to an estimated $1.27 
savings in Medicare reimbursements. With an ambu- 
latory substitution rate of 0.3, the benefit-cost ratio 
increases to 1.504; at 0.7 it drops to 1.035. In all three 
cases, the West shows a benefit-cost ratio less than 
one; the positive savings in the North Central region 
are slightly more than offset by the negative savings 
in the South, while the Northeast exhibits a benefit- 
cost ratio ranging from 3.17 to 4.58. Table 49 also 
shows that for substitution ratios of 0.3, 0.5, or 0.7, the 
PSRO program generates reimbursement savings ex- 
ceeding costs by $39.2, $21.0, or $2.7 million 
respectively. 



TABLE 48 

Maximum Potential Shift of Fixed Costs to Non- Medicare Patients, Hospitals, and Other Payors 

(in Dollars per Day Saved) 













5 
VC 
















4 


Savings 
Shifted 


6 








1 
Average 


2 

(VC) 
Variable 


3 
(FC) 
Fixed 


Sum of 
Variable 

Cost 
Effects 

for 
Routine 


to 

Non- 
Medicare 
Patients 
(Variable 

Cost 


Sum 
Real- 
location 
Effects 

for 
Routine 


Maximum Amount of 

FC Shifted to 

Non-Medicare Patients 

(Col. 6— Col. 5) 




Dollars 


% of 


Census 


Daily 


Cost 


Cost 


& 


Purchase 


and 


Per Day 


Total 


Region 


Per Diem 


Per Diem 


Per Diem 


Ancillary 


Effects) 


Ancillary 


Saved 


Per Diem 


NE 


$192.86 


$76.94 


$115.42 


$37.56 


$39.38 


$113.42 


$74.04 


38.5% 


NC 


183.06 


73.22 


109.84 


31.74 


41.48 


117.32 


75.84 


41.43% 


S 


155.50 


62.20 


93.30 


—26.42 


—35.78^ 


—99.72 


—63.94 


—41.12% 


W 


243.42 


97.37 


146.05 


42.06 


55.31 


148.74 


93.43 


38.38% 


Nation 


188.79 


75.52 


113.27 


41.45 


34.07 


125.67 


91.60 


48.52% 



TABLE 49 

Benefit-Cost Ratios for Tliree Alternative Assumptions Concerning 
Ambulatory Substitution Effects 



Census Region 


Sa=0.3 




S^=0.5 




S^=0.7 


Northeast 


4.58 






3.88 






3.17 


North Central 


2.13 






1.79 






1.44 


South 


—2.28 






-1.91 






—1.54 


West 


0.87 






0.72 






0.57 


Nation: B/C ratio: 


1.504 






1.269 






1.035 


Benefits: 


$116.9 


mill 


ion 


$98.7 


m 


llion 


$80.4 million 


Costs: 


77.7 


mill 


ion 


77.7 


m 


llion 


77,7 million 


Benefits 
















Less Costs: 


39.2 


mill 


ion 


21.0 


m 


llion 


2.7 million 



88 



4. Overview of PSRO Review Costs 



4.3 Findings 



4.1 Objectives 

Estimates of the concurrent review costs for IVIedi- 
care discfiarges during CY 1978 were presented as part 
of the benefit-cost analyses (Section 3). In this chapter, 
the discussion of PSRO costs is broadened to include 
all three types of hospital review: concurrent, MCE, 
and areawide. The cost data presented apply to all 
Federal discharges rather than to IVledicare alone. The 
purpose of the chapter is to examine the costs per 
discharge for the different types of review and the 
summed total cost of review. These costs are com- 
pared with the $8.70 unit cost objective set in FY 
1979. In addition to describing the Jevel of costs, this 
chapter will explore the variation in review costs per 
discharge with respect to delegation status, size of 
PSRO (number of discharges), and number of months 
active. 



4.2 Data Sources and Cost Definitions 

The data source for the description of review costs 
is the Unit Cost Report prepared quarterly by HSQB. 
These data are in turn derived from the PMIS Quar- 
terly PSRO Cost Summary (HCFA 151) and the Dele- 
gated Hospital Cost Report (HCFA 153). In the Unit 
Cost Report, data are presented for each cost category 
shown in Table 50. Total review costs are given by 
delegation status for each of three types of hospital- 
based review, and management and support costs are 
shown for each PSRO. Long-term care and ambulatory 
care review costs are not included in the Unit Cost 
Report, nor are they considered in this evaluation. 
The total costs in each category of Table 50 are 
divided by the number of discharges to give the 
review cost per discharge which is referred to as the 
"Unit Cost" in the HSQB Report. The costs per dis- 
charge in each review category for all active PSROs 
are the basis of the descriptive statistics presented in 
this chapter. 

While cost per discharge adjusts total review costs 
for PSRO size, it gives no information on the costs 
per review. This is especially true after 1978 because, 
as focused review has been implemented in more and 
more PSROs, certain categories of patients have been 
omitted from review. There may be a considerable 
difference between the cost per discharge and the cost 
per review in a given PSRO. Since there are presently 
no data collected on the extent of focusing, and no 
comprehensive information on the number of reviews 
being conducted by hospitals or PSROs, it is impos- 
sible to draw inferences on the cost per review. This 
is a basic data limitation which precludes the study 
of the efficiency of the review process. 



Table 51 describes the magnitude and variation in 
the different review cost components for the nation 
and for each of four Census regions. Given the nature 
of the distributions and the small number of active 
PSROs in each region, the median cost per discharge 
(rather than the mean) is the primary statistic dis- 
played in Table 51 and most of the remaining tables 
in this chapter. 

The median hospital review cost per discharge was 
$12.91 for the nation in CY 1978 compared with $12.31 
in FY 1977 as reported last year. However, while the 
total review cost rose slightly, the unit cost for con- 
current review alone fell to $8.57 from the $8.76 
median reported last year. Increases appear in area- 
wide review costs which rose from $2.21 last year to 
$3.24 in CY 1978. For most cost components, the unit 
costs are lowest in the North Central region and high- 
est in the West. The West has by far the widest dis- 
persion of costs as is evident from the relatively large 
standard deviations (in parentheses. Table 51). Similar- 
ly, the large variation of unit costs for each compo- 
nent is clear from the minimum and maximum values 
reported at the national level. 

One should guard against the inference that areas 
with low costs per discharge have low costs per re- 
view or are somehow more efficient. In fact, cost per 
discharge may be low because only a small number 
of discharges are subjected to review. If a large num- 
ber of discharges are focused out, a PSRO could have 
a low cost per discharge and, at the same time, a high 
cost per review.i 

Analyzing the PSRO review costs, as above, by ex- 
amining the review costs per discharge is one way to 
lool< at the variation in review costs and relative effi- 
ciency of review among PSROs. This constitutes a 
study of PSRO review efficiency in a gross sense so 
long as one defines hospital output as the number of 
discharges. More appropriately, hospital output may 
be defined as the total days of care provided with a 
given quality. A study of the review efficiency among 
PSROs with this definition would require data on the 
variation in review costs for a given impact on days 
used with a given quality of care. By this strict defini- 
tion of hospital output, neither data on cost per dis- 
charge nor costs per review are adequate to examine 
review efficiency questions without information on the 
impact of the review process on both utilization and 
quality. 

The distribution of total review cost per discharge, 
i.e., the sum of concurrent, MCE, and areawide review 
costs, is displayed in Table 52. The total review cost 
per discharge varies widely among PSROs; 65 per- 
cent fall between $10.00 and $16.00. The median of 
the distribution is $12.91. There is no evidence to sug- 
gest that the total review costs are approaching the 



' As noted earlier, there are no data on the number of 
reviews conducted. Therefore, the cost per review cannot 
be calculated. 



89 



TABLE 50 
Cost Data Included in Unit Cost Report 



Cost Category 



Definition of Costs 



PSRO Management and Support Costs 

Total Hospital Review Costs: 
— Areawide Review Costs 



-Medical Care Evaluation Review Costs: 
Delegated MCE review costs 
Non-delegated MCE review costs 

-Concurrent Review Costs: 

Delegated concurrent review costs 
Non-delegated concurrent review costs 



Includes all PSRO personnel costs, consulting and 
accounting services, and all non-personnel costs such 
as travel, rent, furniture, equipment, etc. 

Sum of areawide, MCE, and concurrent review costs. 

Includes costs of patient, practitioner, and institutional 
profiles, data processing, criteria development, hospital 
review supervision and clerical support for these 
functions. 

Includes costs of assessing the quality of delivery and 
organization of health care services. These duties 
may be non-delegated (performed by the PSRO) or 
delegated to the individual hospitals. 

Includes the costs of conducting initial admission 
certification and continued stay review. Concurrent 
review may be delegated to hospitals, or non- 
delegated (performed by the PSRO). 



TABLE 51 
Summary of 1978 PSRO Review Cost Components 



r^PHQI IQ 


Review 


Cost Per Discharge 




Total 

Review 

Cost 


Management 

and 

Support 

Costs 


Region 


Concurrent 
Review 


MCE 
Review 


Areawide 
Review 


Northeast 


$8.56* 


$1.16 


$3.35 


$13.59*** 


$5.07 




(2.34)** 


(.579) 


(1.24) 


(2.61) 


(2.76) 


North Central 


6.35 


1.26 


2.38 


10.88 


4.61 




(2.29) 


(.713) 


(.938) 


(2.43) 


(2.51) 


South 


8.04 


0.89 


2.41 


11.48 


5.69 




(2.33) 


(.589) 


(1.17) 


(2.94) 


(3.92) 


West 


9.96 


1.20 


4.36 


15.72 


6.50 




(4.38) 


(.961) 


(2.93) 


(6.03) 


(13.80) 


National: 












Median 


$8.57 


$1.19 


$3.24 


$12.91 


$5.33 


Mean 


8.81 


1.28 


3.61 


13.68 


7.10 


Std. Dev. 


(3.25) 


(.736) 


(2.01) 


(4.43) 


(7.88) 


Maximum 


21.68 


4.01 


14.13 


30.74 


79.60 


Minimum 


2.75 





0.01 


6.92 


1.13 



Median cost per discharge in dollars. 

Numbers in parentheses are standard deviations. 



Note that medians cannot be summed; thus, concurrent 
review, MCE and areawide review costs do not sum 
directly to total review cost. 



90 



TABLE 52 



TABLE 53 



Distribution of PSROs Among Total Review 
Cost Per Discharge Categories* 

Cost No. Cost No. Cost No. 

Range PSROs Range PSROs Range PSROs 



Distribution of PSROs Among Concurrent Review Cost 
Per Discharge Categories by Delegation Status 



$ 6-8 
$ 8-9 
$ 9-10 
$10-11 



3 

7 

9 

15 



$11-12 
$12-13 
$13-14 
$14-15 



16 
10 
12 
14 



$15-16 
$16-20 
$20-25 
$25-31 



11 

10 

8 

3 



Median $12.91 
Mean $13.68 
* Excludes management and support costs. 

$8.70 target figure designated in 1979. While 17 per- 
cent of active PSROs had total review costs below 
$9.00 in 1977, Table 52 shows that only 8.5 percent 
(10 of 118) met this criterion in CY 1978. This suggests 
that much more focusing is needed to reduce the 
number of reviews and lower the total review ex- 
penditures in order to reduce the cost per discharge 
to the target level. 

One might hypothesize that the total review costs 
may be high due to the influence of start-up costs in 
newly implemented PSROs. However, since this data 
base has only a few "young" PSROs, and since data 
from the PSRO Evaluation have demonstrated that 
newly activated PSROs attain 80 percent or more im- 
plementation in four to six months, it is unlikely that 
start-up costs are responsible for the magnitude of 
review costs shown in Table 52. 

In some cases concurrent review is delegated to the 
hospitals, and in others it is conducted by the PSROs. 
It is logical to suspect that differences in delegation 
status may be associated with differences in review 
costs. The distribution of concurrent review costs are 
shown in Table 53 for all reviews, delegated reviews 
only, and non-delegated reviews only. The median con- 
current cost for non-delegated review alone is $10.71, 
considerably higher than the $7.59 for delegated re- 
view. Overall, the median cost of review for both dele- 
gated and non-delegated review is $8.57. For 
delegated review, 60 percent of the PSROs have 
concurrent review costs between $5.00-9.00, whereas 
only 24 percent of PSROs with non-delegated review 
are between $5.00-9.00. In general, the patterns in 
concurrent cost between delegated and non-delegated 
review are typical of the findings in previous 
evaluations. It should be noted that while delegated 
and non-delegated review costs are compared above, 
these costs are not strictly comparable due to 
differences among hospitals in the allocation of 
overhead costs. 

It should again be stressed that the data of Table 
53 do not imply that delegated review is in any way 
better or cheaper than non-delegated review. In fact, 
the probability of being delegated increases with 
hospital size, and large hospitals are also more likely 
to focus out a large percentage of admissions than 



Cost 
Range 



Total 



Delegated 



Non- 
delegated 



zero activity 

$1-3 

$3-5 

$5-7 

$7-9 

$9-11 
$11-13 
$13-15 
$15-17 
$17+ 



1 

6 

32 

30 

26 

10 

5 

4 

4 



3 

4 

7 

38 

32 

18 

11 



2 

3 



32 

3 

2 

10 

10 

19 

6 

7 

7 

22 



Median* 
Mean* 



$8.57 
8.81 



$7.59 
7.99 



$10.71 
16.89 



* Excludes cases where costs are zero. 



smaller hospitals. These factors could easily be re- 
sponsible for the differences between delegated and 
non-delegated concurrent review cost per discharge. 
If one's objective were to evaluate the cost-effective- 
ness of delegated versus non-delegated review, de- 
tailed data on cost per review would be required. 

Another factor thought to be associated with the 
magnitude of review costs is the size of the PSRO as 
measured by the number of discharges subject to 
review. One would suspect that the cost per discharge 
would be greater in the smaller hospitals and PSROs 
as they have fewer admissions over which to dis- 
tribute the fixed costs of concurrent review. However, 
the data do not support this hypothesis. 

Table 54 shows the median concurrent review costs 
per discharge for three size-of-PSRO categories for 
all reviews, delegated review, and non-delegated re- 
view. The smaller PSROs do not seem to have higher 
concurrent review costs for delegated review, but do 
show unusually high costs for non-delegated review. 
On the other hand, large PSROs do seem to have 
substantially lower costs in the case of delegated re- 
view but not in non-delegated review. As mentioned 
above, issues of efficiency can not be addressed with 
the data of Table 54. Also, comparisons between dele- 
gated and non-delegated concurrent review costs are 
not strictly comparable due to differences among 
hospitals in the way overhead costs are allocated. 

In addition to variations in concurrent review costs 
by size of PSRO and delegation status, one might ex- 
pect the magnitude of review costs to be related to the 
length of activity. The median costs for concurrent 
review, for all reviews, delegated, and non-delegated, 
are shown by length of activity as measured by the 
number of months since the first hospital in the PSRO 
began active review (through July 1, 1978). Since the 
effect of start-up costs would show up primarily in 
management and support costs, these costs are not 



91 



TABLE 54 



TABLE 55 



Median Concurrent Review Cost by Size 
of PSRO and Delegation Status 



No. of 
Discharges 



0-29,999 



30,000-49,999 



50,000-415,000 



Total 



Delegated 



Non- 
delegated 



$8.43 
(38)^ 



$7.59 
(37) 



$13.53 
(25) 



8.72 
(35) 



8.58 
(35) 



10.22 
(24) 



8.19 
(45) 



$6.93 
(45) 



$10.56 
(37) 



* Number of cases included in the calculation. Cases where 
costs are zero are excluded. 



Median Concurrent Review Costs by Months by Review 
Activity and Delegation Status 



No. of 

Months 

of Review 


Total 


Delegated 


Non- 
delegated 


0-20 


$7.82 
(37)* 


$7.59 
(37) 


$11.53 
(26) 


21-30 


8.56 
(37) 


7.93 
(37) 


14.22 
(27) 


31-50 


8.73 
(44) 


7.51 
(41) 


10.56 
(33) 



' Number of cases included in the calculation. Cases where 
costs are zero are not included. 



reflected In Table 55. The unit costs for PSROs in the 
first 20 months of review, for delegated and non- 
delegated review, are not substantially different from 
the costs for older PSROs (31-50 months). However, 
concurrent review costs are considerably higher for 
PSROs of intermediate age (21-30 months) especially 
for non-delegated review. This is an anomaly in the 
data that is not easily explained. 
While neither Tables 54 or 55 show higher costs for 



smaller and younger PSROs as one might expect, both 
tables highlight the large cost differences between 
delegated and nondelegated review. 

In summary, the findings this year appear very 
similar to those contained in last year's report. The 
primary finding of this section is that PSRO total re- 
view costs at $12.91 are still considerably greater than 
the target of $8.70 for the end of FY 1979. 



i 



92 



J 



5. Quality Assurance Activities: Medical 
Care Evaluation Studies 



5.1 Objectives 

Throughout the legislative history of PSRO, the 
theme of utilization control has been paralleled by the 
concern that the services delivered meet professional 
standards of quality. While opinions have varied re- 
garding the relative emphases on these two aspects 
of the program, it has always been recognized that 
the reduction of utilization rates at the expense of the 
quality of care provided to Medicare, Medicaid, and 
Maternal and Child Health beneficiaries would be 
unacceptable. Further, the PSRO provisions of the 
Social Security Act imply that PSROs should not only 
maintain quality of care, but actively seek out problems 
and eliminate them. This chapter of the evaluation re- 
port deals with the PSRO program's current principal 
mechanism for monitoring, assuring, and improving 
the quality of medical care: the medical care evalua- 
tion studies (MCE) component. 

After a discussion of the nature of MCEs, their 
place in the PSRO program, and their implementation, 
two studies are presented. The primary study examines 
the issue of MCE impact: Can changes in the quality 
of medical care be measured which are associated 
with the conduct of MCEs? To address this issue, a 
special survey of.PSROs was conducted to obtain 
Information on the degree of compliance with stand- 
ards of good medical care before and after sampled 
MCEs were performed. 

The second study is more exploratory, relating the 
benefits of MCEs, in terms of improved patient health 
status, to two kinds of costs: the costs of conducting 
MCEs, and the changed costs of medical care occa- 
sioned by improved compliance with standards. The 
result is an estimate, though approximate, of the bene- 
fit/cost ratio which can be attributed to MCEs. 

An additional study is in progress at this time and 
will be reported separately. It consists of a series of 
case studies which examine the evolution and impact 
of MCE programs in six PSROs in order to isolate the 
factors which facilitate or hinder PSRO quality assur- 
ance efforts and to validate the findings of the varia- 
tion rate study. 



5.2 Definition of MCEs 

For most PSROs the mandate to monitor and im- 
prove quality of care has been carried out through 
medical care evaluation studies. The procedures of 
medical audit — a method of systematic retrospective 
peer review of case records which has gained increas- 
ing acceptance in recent decades — serve as the basis 
for the MCE process. Generally, medical audits involve 
a series of steps: a committee of providers selects 
an area of care that may be problematic; criteria of 
good care are specified, including standards that each 



case is expected to meet; a sample of records is re- 
viewed according to the specified criteria and stand- 
ards; cases that do not meet standards are reviewed 
by the committee to determine whether the observed 
variations are justified or constitute deficiencies. 

Five characteristics of MCEs conducted under the 
PSRO program, however, make them different from 
traditional audits: 

1 . In the PSRO program, reaudits are required to 
confirm the effectiveness of recommended ac- 
tions in correcting deficiencies; 

2. PSROs are responsible for the number and qual- 
ity of MCEs performed in hospitals; 

3. PSROs facilitate area-wide (multiple-site) audits; 

4. Central and regional offices of the PSRO program 
provide technical assistance and monitoring in an 
attempt to ensure uniform performance of MCE 
activities; and 

5. The MCE process is linked to other PSRO 
mechanisms — e.g., problems detected through 
profile analysis may provide subjects for MCEs, 
and MCE studies may provide feedback for 
concurrent review. 



5.3 Evolution of PSRO Quality Assurance 
Requirements 

Systematic attempts to assure the quality of hospital 
care are usually traced to the work of Groves, Cabot, 
and Codman early in this century, at a time when re- 
forms were also taking place in medical education in 
response to Dr. Abraham Flexner's renowned report 
to the Carnegie Foundation.i These physicians car- 
ried out studies of the processes and outcomes of 
care rendered in their hospitals by reviewing patient 
records and thus are considered to be forerunners of 
the modern medical audit. The first attempt to apply 
such quality standards in a large survey of hospitals 
was in a study by the Clinical Congress of Surgeons 
of North America (which soon thereafter became the 
American College of Surgeons) in 1916. The results of 
this study were so controversial that they were never 
published: of 692 hospitals surveyed, only 89 (13%) 
could meet minimal standards. 



^ The history of quality assurance is reviewed in more detail 
in a number of studies, including: 

R. H. Brook and A. Davies-Avery, Mechanisms tor 
Assuring Quality of U. S. Medical Care Services: Past, 
Present, and Future, Santa Monica: Rand Corporation, 1977. 

P. A. Lembcke, "Evolution of the Medical Audit," Journal 
of the American Medical Association 199 (February 20, 
1967): 111-118. 

J. W. McAllister, "Quality of Medicare Care: Assessment 
and Assurance," National Center for Health Services 
Research, 1978. 

USDHEW, Office of Planning, Evaluation, and Legislation, 
Health Services Administration, PSRO: An Evaluation of the 
Professional Standards Review Organizations (Rockville, 
Maryland: USDHEW, 1977), Vol. 2, Chapter 2; and Vol. 7, 
Chapter 2. 



93 



Attention in the next several decades turned from 
examining the quality of processes and outcomes to 
assuring that hospitals met minimal structural require- 
ments in their organization, staffing, facilities, proce- 
dures, and training The American College of Surgeons 
(ACS) was a major force in this movement from 1918 
to 1952 through its Hospital Standardization program. 
Many of the same standards eventually became em- 
bodied in state hospital licensure statutes. 

In 1952, an organization for hospital accreditation 
was established through the formation of the Joint 
Commission for Accreditation of Hospitals (JCAH) by 
the American College of Surgeons, the American Col- 
lege of Physicians, the American Medical Association, 
and the American Hospital Association. In the begin- 
ning the Joint Commission adopted the ACS quality 
standards, but it has continually revised and extended 
them. One significant addition was the requirement 
that "control committees" be formed among the medi- 
cal staff, such as medical records, tissue, infection 
control, pharmacy, and utilization review committees, 
among others. These committees were to carry out 
systematic continuous reviews of clinical experience 
in the hospital, often on the basis of patient records. 

Since the passage of the Medicare and Medicaid 
amendments in the mid-1960s. Federal and JCAH re- 
quirements for hospital quality assurance have evolved 
in tandem. The earliest regulations specifying condi- 
tions for hospital participation in Medicare called for 
standards similar to those promulgated by the JCAH 
at that time. Between the passage of the PSRO legis- 
lation and its implementation, the JCAH requirements 
became much more specific: the Commission de- 
veloped an audit methodology known as Performance 
Evaluation Procedure (PEP) for auditing and improv- 
ing patient care and sponsored workshops to train 
health professionals in its use. By mid-1975, any hos- 
pital surveyed by the JCAH had to display two com- 
pleted audits of acceptable quality and two in prog- 
ress. At the time that the PSRO program was begin- 
ning quality of care review, therefore, a methodology 
of medical audit had been established nationwide. This 
methodology was adapted by the PSRO program as 
the basis for MCEs. Requirements for the numbers of 
audits to be performed annually were adjusted by both 
JCAH and the PSRO program in 1976-77, ranging 
from four to twelve per hospital depending on the num- 
ber of patients discharged annually. Thus, during the 
period considered in the studies reported in this 
chapter, hospitals were responding to nearly identical 
requirements of JCAH and PSRO, and often reported 
the same audits to both bodies. 

Joint Commission and PSRO quality assurance 
standards continue to evolve, and will be significantly 
different in the 1980s than they have been in the 
1970s, both in their basic philosophy and in the prac- 
tical steps which hospitals will take to satisfy them. 
The JCAH has responded to criticisms of medical audit 
by altering its policy; in the future, the renewal of a 
hospital's accreditation will depend partly on its 
ability to demonstrate that it has maintained a com- 



prehensive quality assurance system. The emphasis 
will be on a hospital's ability to detect and resolve 
problems in care, and not on the completion of a 
specifed number of audits.^^ The effectiveness of this 
policy will depend on how the hospitals respond to the 
new requirements, and on how the JCAH promulgates 
and enforces them. 

The PSRO program has similarly altered its policy, 
de-emphasizing requirements for numbers of quality 
reviews in favor of demonstrations that problems in 
care have been resolved. PSROs will be expected to 
develop lists of problems, to ascertain that activity is 
taking place to resolve them, and to show that these 
actions have produced changes in care. Success in 
these endeavors will be one of the factors reviewed 
in evaluating the performance of each PSRO and in 
considering its applications for renewed funding. 

Quality assurance policies are thus becoming less 
specific. It is difficult to predict what this trend will 
mean for quality of care. In part, the change 
represents a dissatisfaction with "paper compliance" 
with formal requirements. Numbers requirements and 
set methodologies have been useful, at a minimum, in 
involving hospital and PSRO staffs in quality assurance. 
Further progress will depend on their continuing 
commitment to quality, and on the conviction with 
which the Joint Commission and the PSROs monitor 
their accomplishments. 

The studies of MCE impact, benefits, and costs which 
follow are thus related to a finite period in the evolution 
of PSRO quality assurance. They are relevant as an 
evaluation of the quality of care efforts of the PSRO 
program from its implementation to the present day, 
and as a baseline for the measurement of future 
impacts. The case studies of PSRO quality assurance 
efforts will include an examination of how several 
PSROs perceive and intend to implement the new 
policy. These will be reported early in 1980. 



5.4 Development and Current Status of 
MCE Activity 

5.4.1 Objectives 

This section describes the implementation of the 
MCE component of the PSRO program in terms of the 
number of audits and reaudits completed from 1975 
through 1978 and the most frequent MCE topics in 
1978. The characteristics of the MCE cycle are 
specified in PSRO Transmittal No. 43 * and the 



- Joint Commission on Accreditation of Hospitals, "Alternative 
Approaches to the Review of the Quality of Patient Care 
in Hospitals," discussion paper presented to the National 
Professional Standards Review Council, 1979. 

" Joint Commission on Accreditation of Hospitals, 
Accreditation Manual For Hospitals, 1980 Edition (Chicago: 
JCAH, 1979). 

' DHEW, HSA, BQA, PSRO Transmittal No. 43, January 25, 
1977. 



94 



Handbook for the Conduct of Medical Care Evaluation 
Studies.^ 

5.4.2 Data 

Information on MCE activity has been gathered 
through the PMIS since the beginning of the program in 
1975. PSROs submit reports on each MCE audit and 
each reaudit completed, as well as quarterly listings of 
all MCEs, completed and in progress. The first two, 
HSQB Forms 131 and 133, form the basic data source 
for this report. 

PMIS data are maintained by HSQB in computerized 
form. Reports from the PSROs received by the PMIS 
data contractor are checked for completeness and 
consistency before key entry; problems are clarified by 
telephone if possible — otherwise the forms are 
returned to the PSROs for correction. The quality of the 
data reported here was further improved by a 
painstaking comparison of the computer file with PSRO 
reports. The tables thus have a higher degree of validity 
than those reported in previous years, and detailed 
comparisons with previous tables should not be made. 
Data for the period 1975-78 were continuously updated 
during 1979, eliminating many of the problems of 
reporting and processing lags which occurred in 
previous studies. The time reported in the figures and 
tables which follow is the date that the MCE (audit or 
reaudit) was completed by the PSRO or hospital rather 
than the date it was entered into the data base. 

5.4.3 Methods 

MCE topics are selected by committees of physicians 
as an aspect of the MCE design process, and may be 
narrowly specified. For purposes of this analysis, topics 
were grouped into 578 topic categories judged to 
preserve the most meaningful distinctions among 
topics. « These topic categories generally represent 
medical diagnoses or surgical and administrative 
procedures. 

No analysis by hospital is presented here, since 
hospitals are not identified in PMIS reports. Thus 
compliance with MCE numbers requirements cannot be 
assessed. 

5.4.4 Quantity and Distribution of MCE Activity 

Initial Audits. The number of initial MCE audits 
reported each quarter from early 1975 through 1978 is 
shown in Figure 17. The figures show a large increase 
in the number of MCEs during this period. As Figure 18 
shows, this increase is a product both of the greater 
number of PSROs reporting MCE audits, and the larger 
average number of audits per PSRO. 

Some PSROs are far more active in conducting MCEs 
than others. Figure 19 shows the proportion of all 



MCEs done in 1977-78 which were reported by the 
ten most active PSROs, the next ten, and so on. The 
twenty most active PSROs performed about 50 percent 
of all initial audits, the next twenty performed about 20 
percent of the audits, while the other remaining PSROs 
reported only 30 percent of the total. 

Table 56 shows the number of MCE audits reported 
by each PSRO from mid-1975 through 1978. The PSRO 
totals range from one MCE to 1,346 during this period. 

The disparity in MCE productivity is partly explained 
by the varying ages and sizes of the PSROs, since the 
numbers of MCEs reported can be expected to be 
proportional to the length of time a PSRO has been 
conducting review and the number and sizes of its 
hospitals. Yet, examination of Table 56 shows that this 
is not entirely the case: while nine of the ten most 
active PSROs began early and maintained a high level 
of MCE activity, some of the older ones did not, and 
some newer PSROs have become quite active. 

One pattern which can be discerned is that PSROs 
seem to accelerate to a level of MCE activity over a 
few quarters and then maintain that level. Very few 
PSROs have increased and later decreased their 
productivity. The inference is that each PSRO is 
characterized by an implicit or explicit policy regarding 
the level of effort to be expended on MCEs, and that 
this is an important factor in its productivity. It 
remains to be seen how those PSROs which have 
reported few audits (and those which have reported no 
audits and thus are not represented in Table 56) are 
fulfilling their responsibility to monitor and improve the 
quality of care. 

Reaudits. Reaudit frequencies, presented in Figure 
20 by quarter, are also increasing, but are considerably 
lower than the figures for initial audits. Figure 21 shows 
that the increasing number of reaudits reported is partly 
due to a rise in the average number of reaudits per 
reporting PSRO, but that it is primarily a function of the 
increasing numbers of PSROs which report them. As 
shown in Figure 22, however, the twenty PSROs which 
reported the most reaudits account for the majority 
(60%) of reaudit activity, with the other 40 percent 
accounted for by the remaining 81 PSROs. Table 57 
shows the number of reaudits by quarter for each 
reporting PSRO. 

Some lags between audit and reaudit figures may 
be expected due to the fact that reaudits are usually 
performed up to one year after the initial audit, but 
even allowing for this lag, the number of reaudits is 
one-third of the number expected.^ PSROs are thus not 
receiving and/or reporting sufficient feedback on the 
efficacy of actions taken to remedy deficiencies found 
in the initial audits. 

Audit topics. The ten most frequently audited 
medical topics for 1978 are shown in Table 58. These 



^DHEW, HCFA, HSQB, 1978. 

' Coding was carried out according to Equivalence Tables 

5 and 6, published by the Commission on Professional 

and Hospital Activities, 1974. 



'This estimate is based on the finding that 70 percent of 
initial audit reports (HSQB Form 131) indicate that one or 
more unjustified variations were found. The corresponding 
estimate in the 1978 report was one-fifth. 



95 



Figure 17 
MCE Initial Audit Activity By Quarter 



4000 








7a 


1975-76 


3500 


-■ 


19 77 




— 


19 78 



3000 - 



2500 - 



in 
H 

M 

p 



H 
M 

!a 



H 
O 
H 



2000 



1500 - 



1000 - 



500 - 



3664 



3241 



3268 



3088 




2-75 



3^5 4-75 1-76 2-76 3-76 4-76 1-77 2-77 3-774-77 1-78 2-78 3-78 4-78 



QUARTER 



96 



Figure 18 
Average PSRO Audit Activity By Quarter 



Ph 

w 

H 
M 
O 

w 








QUARTER: 
NUMBER OF 
PSROs 
REPORTING : 




2-75 3-75 4-75 1-76 2-76 3-76 4-76 1-77 



10 18 33 42 57 59 71 79 



97 



Figure 19 
Percent of Initial Audit Activity By Ranked Groupings (1977-1978) 



H 

u 

<: 

w 
u 

<: 

H 

o 

H 

Ph 
O 

w 

55 

<i 

w 

CJ 

pi 
w 
P-I 




1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 91- 100- 111- 121- 

100 110 120 141 
RANK GROUPINGS OF PSROs 



98 



I 



TABLE 56 
Quarterly Distribution of MCE Audit Activity 



PSRO 

l.D. 

NUMBER 








OfAP-f 


RS 








1 


TOTAL 

THROUGH 

1978 


1975 
12 3 4 


1 


1976 
2 3 


4 


1 


1977 
2 


3 


4 


1 


1978 
2 3 


4 


CAKOl 





2 1 


3 


6 


1 


6 


2 


14 2 


1 


46 


OIDOl 






1 


5 


5 


2u 


38 


26 


41 


11 14 


24 


189 


Qopo: 


6 


-, 


4 4 


2 


1 1 


IP 


17 


31 


w 


15 21 


21 


168 


OORC 




«s 


14 :i 


36 


48 


66 


48 


On 


37 


9 10 


7 


397 






i 


7 19 


41 


74 


128 


94 


64 


103 


122 199 


235 


1092 




1 1 


- 1 


16 17 


1.=. 


1 


20 


.- 


18 


12 


19 19 


26 


...248 


irTC 




4 


3 8 


b 


13 


12 


19 


L-} 


i: 


11 6 




143 


irTfTl 






15 20 


?4 


16 


20 


25 


29 


TJ 


33 11 


9 


235 


ICTOA 




7 


c 5 


10 


10 


8 


16 


16 





7 2 


9 


104 


INAOl 






2 11 


8 


19 


22 


27 


37 


25 


34 24 


25 


234 ■ 


1NA02 




2 \ 


4 


17 


9 


16 


7 


9 8 


18 


90 


1MA03 


2 8 


8 


12 7 


19 


10 


14 


6 


23 


9 


8 7 


18 


151 


1MA04 


3 


1 


5 11 


6 


43 


98 


62 


67 


59 


84 45 


12 


496 


IM.^05 






3 13 


21 


27 


22 


26 


27 


13 


12 20 


41 


225 


IMEOl 


3 





20 20 


30 


36 


46 


15 


25 


18 


15 14 


7 


256 


INHOl 








8 


3 


1 




12 


1RI02 






2 







12 


16 


29 


31 


32 11 


46 


197 


1\T01 








5 


11 


14 


10 


10 22 


38 


110 


2NJ01 






2 12 


25 


7 


22 


5 


22 


12 


10 13 


21 


149 


2NJ02 


2 


i 


4 14 


9 


15 


10 


13 


6 


12 


16 24 




126 


JK.TOi 




1 


6 


9 


9 


24 


14 


31 13 


23 


130 


2NJC5 










5 


3 


8 


2NJ06 






1 


5 


6 


2NJ08 






2 


5 


5 


1 


13 


2NY01 


2 








2 


2NV02 


2 


24 


23 21 


23 


12 


25 


31 


21 


9 


9 14 


38 


242 


2KY03 






10 


10 


24 28 


47 


119 


2NY0A 








4 


7 


6 


7 


3 14 


13 


54 


2N'Y05 






2 


'J 


6 


16 


10 


25 


i: 


11 


7 


94 


2N'Y06 






1 




13 


1 


15 


2NY09 




4 


15 12 


21 


24 


22 


25 


48 


15 


22 20 


40 


268 


2NY10 


1 


5 


7 3 


6 


4 


2 


6 


1 


1 


4 • 4 


11 


55 


2NT11 






8 4 


17 


24 


28 


19 


68 


16 


57 24 


74 


339 


2m'12 








5 


4 


12 


4 


6 5 


3 


39 


2CT13 






4 4 


3 


9 


29 


23 


29 


32 


51 19 


23 


226 


2NT14 






1 


2 


15 


19 


18 


26 19 


25 


125 


2NY:5 


4 1 


13 


4 


; 


9 


19 


26 


3^i 


19 


24 2- 


39 


222 


2NT16 


5 


10 


7 7 


9 


18 


9 


7 


15 


19 


16 14 




136 


2PR01 








1, 


16 25 


16 


61 


3DC01 


1 


6 


4 7 


4 


21 


25 


20 


19 


1- 


23 12 


21 


177 


3DE01 








3 


9 


10 


9 


7- 8 


7 


53 


3MD02 


2 24 


19 


33 27 


20 


16 


34 


15 


43 


15 


2u 16 


49 


337 


3MD03 


1 


3 


2 


: 


^ 


7 


4 


6 


10 


5 8 


5 


58 


3MD0i 




1 


2 3 








' 


3>a3C5 


8 


u 


17 2 


6 


10 


14 


13 


38 


11 


9 16 


19 


169 


3MD06 


9 


10 


9 8 


5 


t 


11 


7 


5 


10 


6 5 


7 


98 


3KD07 


1 13 


10 


11 1 


4 


8 


10 


1 




13 


13 13 


11 


109 


3PA01 








: 


7 


• 3PA02 




2 


11 


24 


12 


16 


19 


2- 15 


34 


159 


3PA03 








2 


2 


3PA04 










25 


15 


17 


17 14 


35 


123 


3PA05 










9 22 


23 


54 


3PA0f. 




3 


13 33 


49 


47 


69 


41 


59 


-3 


52 51 


51 


Ml 



99 



TABLE 56 (continued) 
Quarterly Distribution of IVICE Audit Activity 



PSRO "" 
I.D. 

NUMBER 


1975 
12 3 4. 


1 


1976 
2 3 


OUAATERS 
4 1 


1977 
2 3 


4 


1 


1978 
2 3 


4 


TOTAL 

THROUGH 

1978 


3PA07 




3 


13 21 


19 


23 


26 


30 


16 


19 


29 33 


20 


257 


3PA08 




1 


3 


6 


9 


28 


18 


19 20 


22 


125 


3PA09 


1 


18 


35 39 


59 


66 


55 


55 


70 


46 


75 57 


52 


623 


3PA11 


11 


3 


17 29 


46 


30 


41 


45 


30 


23 


38 28 


21 


367 


3PA12 






2 


21 


31 


43 


9 


43 


46 


10 44 


77 


326 


3VA01 






5 


1 


12 


3VA02 






4 


18 


14 16 




52 


3VA03 








12 1 


12 


3VA04 










1 


12 


13 


3VA05 






1 


13 


8 


7 


21 


24 22 


22 


L13 


3WV01 


3 


6 


12 26 


9 


45 


76 


42 


60 


61 


68 40 


44 


492 


4AL01 






1 1 


26 


46 


77 


105 


130 


141 


137 139 


292 


1095 


4n,03 




1 1 


1 


20 


29 9 


41 


100 


4n.i2 






1 2 


19 


51 


35 


47 


» 


31 


57 45 


57 


400 


4KT01 




1 


1 




21 


35 


56 


4HS01 


24 36 


44 


56 52 


90 


133 


127 


128 


182 


108 


133 129 


104 


1346 


4NC01 










1 


3 


4 


4NC02 








1 


13 13 


26 


53 


4NC03 










2 


18 


20 


4NC04 








1 


2 3 




6 


4NC07 










8 13 


2 


23 


4NC08 






1 


1 


2 2 


9 


15 


4SC01 


2 8 


16 


19 70 


64 


77 


60 


14 


63 


74 


73 75 


61 


676 


4TN01 






1 




1 


4TN02 


2 17 


29 


64 74 


109 


97 


113 


148 


131 


148 


171 129 


7 


1239 


51L01 




1 


20 


20 


5IL02 








5 


5 


5IL03 








63 


84 


203 


127 


150 156 


148 


931 


5IL04 


1 4 


3 


2 2 


5 


3 


10 


4 


7 


8 


11 3 


12 


75 


5U05 






17 


1 






1 


12 33 


51 


115 


51108 








1 


1 


5IN02 








5 


5 


5TS04 










2 


3 


5 


5IN05 










1 14 


12 


27 


5IN06 








5 


5 


5M101 




3 




1 


1 


L 




7 


2 


15 


5M102 






10 




10 


SMins 


2 


3 


9 17 


8 




19 


10 


15 


" 


7 8 


25 


.136 


5K107 










3 


11 


14 


5MN01 






3 


1 




1 




1 


7 


4 6 


45 


68 


5MN02 


7 19 


25 


25 23 


44 


33 


31 


35 


23 


24 


33 27 


48 


402 


5MN03 










2 


13 


11 


17 27 


2 


72 


50H01 


2 


2 


28 57 


31 


36 


48 


41 


38 


26 


32 54 


36 


429 


50H02 




1 


5 


5 


20 


34 


14 


25 34 


22 


160 


50H03 








1 


2 


1 




2 7 


6 


19 


501104 


2 


14 


11 16 


24 


25 


27 


27 


45 


27 


39 25 


27 


309 


50H05 








1 


1 


50H06 










3 


12 


15 


50H10 






11 


25 


31 


41 


16 


20 




144 


50HH 










37 


30 


;o 


50H12 








6 


14 


11 


6 


29 




66 


50H13 










41 


4^ 


S4 


5U101 


1 


1 


5 28 


55 


113 


122 


58 


45 


69 


108 110 


21 


?3« ■ 



100 



TABLE 56 (continued) 
Quarterly Distribution of MCE Audit Activity 



■' 


PSKO 

I.D. 

HLTMEER 


1975 
L 2 3' 4 


1 


1976 
2 3 


QUARTEHS 
4 1 


1977 
2 


3 


4 


1 


1978 
2 3 


4 


TOTAL 

THROUGH 

1978 




5U102 


4 24 


43 


49 46 


56 


39 


50 


53 


79 


32 


59 135 


85 


754 








9 


10 30 


34 


94 


71 


64 


72 


57 


74 65 


92 


555 




flitTuAl 


4 


3 


8 21 


17 


44 


29 


10 


28 


23 


22 24 


22 


260 










25 


55 


57 


59 


84 


101 69 


52 


513 










1 




14 33 


55 


102 




7K002 




7 


29 


36 


23 


58 


9 


31 36 


15 


249 




7M003 






7 


3 


21 


33 


39 


35 


35 


49 36 


43 


303 




TMOO'i 






11 


45 


40 


22 


26 


32 20 


23 


219 




7H005 






9 


18 


18 


18 


24 


23 39 


23 


172 




8C001 






18 35 


49 


48 


58 


46 


66 


73 


62 80 


128 


663 




8MT01 




4 


6 2 


7 


14 


24 


8 


25 


5 


21 16 


5 


137 




8ND01 






23 


U 


30 


53 


56 


39 32 


33 


277 




8SD01 




9 


IS 


32 


30 


22 


31 


22 9 




170 




SUTOl 


2 






3 


2 




7 




8NY01 








15 


1 1 




17 




9CA01 


11 17 


19 


16 3 


' 


7 


5 


9 


^^ 


18 


20 15 


34 


201 




<}r:An2 




1 




' 


5 




9CA03 


3 10 


9 


7 7 


^ 


2 


10 


4 


^ 


17 


13 12 


15 


115 




9CA04 








8 


1 3 




12 




9rA05 






3 


11 


6 


12 


13 


27 34 


24 


135 




QrAOR 


2 


4 


5 2 


14 


14 


L2 


7 


4 


7 


9 16 


6 


102 




9CA09 






2 


3 


5 


5 


13 


25 20 


13 


97 




9CA10 






6 


7 


2L 21 


30 


85 




9CA11 










5 


1 


6 




9CA12 


. 2 


2 


6 18 


13 


3 


7 


5 


1 


1 


2 


5 


70 




9CA13 








1 


4 


6 


11 




9CA14 






2 


15 


29 


22 


20 


21 


14 


25 10 


5 


i6i 




9CA16 










2 


5 


1 


13 17 


5 


43 




9CA17 






1 


1 


3 


3 


1 




10 


13 11 


25 


74 




9CA20 






3 


14 


27 


28 


35 


44 44 


43 


238 




9CA21 








7 


4 17 


5 


34 




9CA22 








17 


5 




2 


24 




9CA23 








4 


5 




38 


42 60 


72 


221 




9CA27 


1 3 


9 


13 4 


9 


17 


23 


10 


3 


7 


15 3 


29 


146 




9CA28 








1 


1 




9H101. 






1 


1 




8 


37 


4 


15 


24 15 


20 


136 




t-NVjl 






5 


19 


21 


4 


22 


19 19 


26 


135 






1 












TOTAL 


80 254 


442 


713 959 


1354 


1923 


2604 


2409 


3088 


2646 


3241 3268 


3664 


26,645 


















1 





















































































































































































101 



Figure 20 
MCE Reaudit Activity By Quarter 



700 




102 





Figure 21 




Average PSRO Reaudit Activity, By Quarter 


Va 


19 75-19 76 


-■ 


1977 


^ 


19 78 


AVERAGE TOTAL // OF INITIAL 


_ PSRO 


= REAUDITS IN QUARTER 


ACTIVITY NUMBER OF PSROs 



o 
Pd 

Pi 

CO 

H 
M 
P 

W 4 

o 



> 



3 - 





QUARTER: 



!.7 



6.5 




6.4 



t.3 



5.4 



2-75 3-75 4-75 1-76 2-76 3-76 4-76 1-77 2-77 3-77 4-77 1-78 2-78 3-78 4-78 



NUMBER OF 
PSROs 
REPORTING : 



1 3 7 10 21 21 26 44 48 57 59 72 81 73 71 



103 



Figure 22 
Percent of Total Reaudit Activity By Ranked Groupings (1977-1978) 



H 
M 
Q 

W 

<: 

H 

o 

H 

O 
W 

H 

M 
O 

W 




1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 91-101 

RANK GROUPINGS OF PSROs 



104 



TABLE 57 
Quarterly Distribution of MCE Reaudit Activity 



psr.o 

I.D. 
NUrtSER 


1975 . 

L 2 3 4 


1976 
1 2 3 


IJUARTt 
4 


R3 
1 


1977 
2 


3 


4 


1 


1973 
2 3 


— i: 


TOTAL 

THROUGH 

1978 




OAKOl 






2 1 


1 


3 




OIDOl 








4 


2 


li 


5 


1 6 


6 


:o 






6-5 2 5 


- 


1. 




I 


4 


5 1 


48 




OORO"" 




1 


: 




1 


1 


3 


9 




OWAOl 




1 1 


'■* 


1 




2 


2 


8 


8 8 


5 


42 




ICTOl 




2 I I 


5 


4 




2 


15 


3 


9 7 


14 


•58 




ICT02 




4 


2 




2 




1 


2 5 




21 




1CT03 






2 


3 


5 


4 




7 11 


2 


34 




ICT04 




2 


1 1 


2 




2 


1 


1 


3 1 


4 



19 




INAOl 




1 


7 


2 


3 9 


1 


28 




1MA02 






3 


1 


1 4 


2 


11 




1MA03 




2 


^ 


4 




3 




4 


1 3 


4 


23 




IMA04 






i 


4 


7 6 


1 


23 




IMA05 




~ 




: 


2 


2 


6 


2 5 


1 


20 




IMEOl 




1 


^ 


4 


4 


2 


6 


3 L 


22 




IRIOl 






1 


5 


1 1 


^ 


12 




tVTfll 








5 3 


8 




2NJ01 




1 


3 


4 




2NJ02 




;; 




6 


2 


1 


4 


2 3 


5 


25 




?«i.!n4 








1 


2 3 


1 


7 




2NJ05 






L 


3 


4 




2MT1-02 




1 


2 


S 


5 


7 


6 


9 


9 4 


4 


57 




2NY03 










1 1 


6 


a 




2NY04 










1 


2 




3 




2>(Y09 






2 


1 


2 


3 


2 


6 5 


12 


33 




2NY10 




1 


1 


3 


2 4 


1 


12 




2SY11 










2 


8 




12 2 


11 


35 




2NY12 






1 


I 


2 




2NY13 






3 


5 


4 1 


8 


21 




2MY14 










1 


3 


5 


13 


1 


23 




2NY15 


1 


3 


1 


1 


3 


15 


10 


4 8 


14 


60 




2>fV16 


1 


1 3 


4 


3 


2 


1 


4 


1 


;o 




2PS01 










1 


11 


12 


1 


3DC01 




2 






2 


11 




4 3 




25 




3DE01 








1 


1 2 




9 




3MD02 




3 2 


5 


4 


7 


7 


6 


7 


11 6 




nT 




3^^D0J 










1 




2 




3MD05 




1 


2 


4 


4 


2 


3 


4 1 




21 




3MD06 






1 


3 


1 


5 


1 


3 2 




17 




3MD07 






2 


2 


1 3 




9 




3PA01 








9 


9 




3P^02 










2 


4 


7 


4 2 




19 




3PA0i 








2 


7 


3 


12 




3PA05 








1 


1 




3PA06 




1 


1 




3 


3 


4 


4 11 


13 


.0 




3PA07 








1 


L 


1 


2 


2 




? 




3PA08 








3 


1 2 


4 


10 




3PA09 




1 2 


3 


3 


a 


7 


3 


11 


12 10 


10 


;o 




3PAil 




L 


2 


7 


6 


3 


7 


11 


3 U 


9 


65 




3PAi2 




1 


1 


3 


1 


13 




3 


9 


31 




3VA0L 




1 1 


1 






3 




3VA02 










2 


1 


1 


1 


6 


11 




3VA03 








3 


^ 





105 



TABLE 57 (Continued) 
Quarterly Distribution of MCE Reaudit Activity 



■ 


fSRO 

I.D. 


1975 
12 3 4 


C'-.-iTrKS 
1976 1977 

12 3 4 12 3 4 


> 

197S 

12 3 4 


rOTAL 
THROUGH 




— 


3VA05 






1 3 


4 4 3 8 


23 




— 


3^01 






4 4 I 


2 3 6 5 


30 




*■ 








4 1 8 12 


1; 15 32 35 


122 




" 


4FT 1 f 




1 1 3 


4 4 22 24 


13 19 15 20 


i:*. 




" 




1 1 


3 5 11 5 


9 6 17 8 


7 4 6 9 


95 






4SC0V 




2 


26 42 25 55 


68 96 58 • 45 


417 




' 


4TtJ02 




1 1 2 


34 34 29 16 




117 






51L02 






5 




5 




■ 


51L03 








10 16 21 30 


77 




511,04 




1 






1 






5M101 








1 1 


2 






5C.105 






14 


1 1 2 


18 






5^C<02 


1 


1 3 2 


4 18 20 16 


11 10 11 7 


104 






5VI1-03 








1 2 


3 






50H01 




2 8 


2 9 8 3 


8 3 2 11 


61 






^0H02 




1 


1 1 3 


3 2 5 6 


22 






50U04 




1 3 


3 5 2 9 


6 13 8 22 


72 






50H10 






6 


6 8 20 


40 






50H12 








12 6 2 


11 






SUlOl 


1 


1 


6 1 1 14 


6 11 5 8 


54 






SW102 




12 2 6 


2 3 5 7 


4 9 12 14 


67 






6i>J(01 






2 5 15 3 


U 20 23 33 


112 






60K01 






3 2 


4 4 


13 






7IA01 






2 


3 15 24 24 


63 






7M002 








2 6 2 2 


12 






7M003 








4 5 4 8 


21 






..7>;o.o.^ 








4 1 1 


6 






,7.^005 






1 3 2 


6 8 3 6 


29 






Rrnni 






18 6 3 


19 7 24 


68 






SKl'Ol 






1 1 2 1 


1 5 1 


12 






8SM1 






2 1 


6 9 5 


23 






8SD01 






2 2 2 


3 1 


10 






3CT01 


6 








6 






9.«,202 


1 








1 






9CA01 




2 4 1 


4 3 2 7 


1 2 4 2> 


32 






9CA03 




2 1 3 


1 1 


2 4 3 7 


24 






9CA04 










1 






9a05 






4 2 2 


4 6 11 7 


36 






9CA03 






2 


1111 


6 






9CA09 






2 2 


1 I ■ 2 


3 






9CA12 






I 6 




7 






9CA,14 








2 1 I 


X 






JCA.i7 






I 3 




4 






9CA20 








3 1 9 


L3 






9C-\-'2 




• 


1 


I 


2 






9ca;3 








9 9 6 19 


43 






9CA27 


* 


1 - 6 


14 3 7 


3 i 3 6 


i.4 






9Ca:9 








1 


L 






9H101 










1 






9:ivoi 






I 4 


< 7 1 9 


".3 






















TCTAl 


1 13 U 


\1 40 54 76 


'.70 -.w 163 :d5 


391 5:1 iol 616 


j-s: 





















106 



TABLE 58 
The Ten Most Frequently Audited Medical Topic Categories, 1978 by Quarter 



Ranki 


ng Topic Category 


1 


Quarter 
2 


3 


4 


Total 


1 


l\/ledicinal Agents 


163 


178 


226 


262 


829 


2 


Pneumonia 


102 


100 


95 


109 


406 


3 


Myocardial Infarction 


95 


103 


100 


90 


388 


4 


Cerebrovascular Accident 


70 


74 


71 


86 


301 


5 


Diabetes Mellitus 


59 


81 


55 


72 


267 


6 


Congestive Heart Failure 


39 


65 


73 


77 


254 


7 


Gastroenteritis 


47 


60 


54 


56 


217 


8 


Urinary Tract Infection 


30 


48 


40 


65 


183 


9 


Hypertension 


33 


38 


27 


56 


154 


10 


Pulmonary Embolism 
TOTAL 


34 


31 


44 


39 


148 




672 


778 


785 


912 


3147 



TABLE 59 
The Ten Most Frequently Audited Surgical Topic Categories, 1978 by Quarter 



Ranking 


Topic Category 


1 


2 


Quarter 

3 


4 


Total 


1 
2 


Hysterectomy, Abdominal 
Cesarean Section 


59 
68 


63 
80 


109 
58 


84 
79 


315 
285 


3 

4 


Cholecystectomy 
Hip Fracture 


68 
49 


67 
65 


63 
60 


77 
59 


275 
233 


5 
6 

7 


Appendectomy 
Tonsillectomy & 

Adenoidectomy 
Normal Deliveries 


59 
50 

48 


57 
68 

71 


51 
47 

45 


51 
44 

38 


218 
209 

202 


8 

9 

10 


Herniorraphy 
Dilation & Curettage 
Cancer of Breast 


30 
19 
13 


26 
25 
36 


33 
34 
20 


37 
44 
36 


126 
122 
105 


10 


Cataract Extraction 
TOTAL 


34 


14 


23 


34 


105 




497 


572 


543 


583 


2195 



are the same topics which were found to be most 
frequent in the 1978 evaluation. The ten most frequent 
medical topics account this year for 25 percent of all 
MCE audits. 

The surgical topics most frequently audited in 1978 
are presented in Table 59. Eleven topics rather than 
ten are included because of one tied rank. Compared 
to the 1977 rankings reported last year, the list is 
similiar, but transurethral resection and cancer of the 
colon do not appear for 1978. In addition, 
herniorrhaphy, dilation and curettage, and cancer of 
the breast have joined the list this year. The ten most 
frequent surgical topics (dropping one of the topics 
tied for tenth place) accounted for 16 percent of all 
MCE audits in 1978. 

Reaudit topics. The ten most frequently reaudited 
topics are shown in Table 60. The most frequent topic 



TABLE 60 

The Ten Most Frequently Reaudited Topic 
Categories, 1978 



Ranking 



Topic Category 



Number of 
Reaudits 



1 


Admissions 


171 


2 


Medicinal Agents 


142 


3 


Myocardial Infarction 


94 


4 


Pneumonia 


77 


5 


Diabetes Mellitus 


45 


6 


Cesarean Section 


43 


7 


Normal Deliveries 


42 


8 


Gastroenteritis 


40 


9 


Cholecystectomy 


40 


10 


Appendectomy 


37 



107 



is an admistrative category which focuses on admis- 
sions, discharges, and transfers. This reflects a use of 
MCEs to investigate topics which are related to efficient 
use of resources. The four surgical and five medical 
topics are also represented on the lists of most 
frequently audited topics. 



5.5 Variation Rate Study 



5.5.2 Study Design and Data Collection 

In contrast to most of the studies reported in this 
volume, the MCE variation rate study required 
collection of data which are not routinely reported, but 
reside primarily at the hospital level. The design and 
methods of this study therefore require a somewhat 
more extensive explanation. This section discusses the 
design and selection of the sample, and the process 
and results of data collection. 



5.5.1 Objectives 

The variation rate study examines a sample of data 
from MCEs to see whether conformance to standards 
of good medical care and patient outcomes improved 
between audit and reaudit. For a sample representing a 
defined population of MCEs, data were gathered on 
the degree to which hospital care conformed to the 
standards used in these MCEs at audit and at reaudit. 
The standards of care studied are those established by 
hospital and PSRO MCE committees when they design 
an MCE. 

The measure of conformance used in this study is the 
variation rate (VR), which is the proportion of patient 
records audited on a criterion (or dimension) of care 
which did not meet the standard set for that criterion. 
The measure of change is the difference in variation 
rate between the time of audit and reaudit. Decreases 
in VR imply improvement in measured performance, 
while increases indicate a decline in measured 
performance. Aggregate variation rates and VR 
changes are calculated when the same criterion of 
care is used in several studies at different sites; these 
aggregates represent the average rate of the average 
change for all audits which use that criterion. 

This study builds upon pilot studies reported in prior 
PSRO evaluations, which established the variation 
rate as a viable measure of quality-of-care changes.^ 
It is the first study to investigate such changes in a 
representative sample of a population of MCEs, not 
simply those which were readily available (i.e., that 
minority of MCEs in which both audit and reaudit had 
been completed). Where reaudits corresponding to 
sampled audits had not been performed, PSROs were 
asked to do them especially for this study. 

Specifically, the study objectives are: 

• to measure changes in VRs from audit to reaudit 
for each selected criterion of care; 

• to measure VR changes for types of criteria, 
particularly those which are judged to have a 
strong or weak relationship with patient outcomes, 
those which increase the cost of care, and those 
which relate most closely with diagnosis, therapy, 
and outcomes; and 

• to measure VR changes for each diagnosis or 
procedure studied. 



' G. S. Adier and A. Dobson, "The Strategy of Evaluating 
Nationwide MCE Impact," Quality Review Bulletin 10 
(October, 1979): 8-11. 



5.5.2.1 SPECIFICATION OF THE PSROs, MCE 

TOPICS, AND CRITERIA TO BE STUDIED 

While it may appear desirable to sample all reported 
MCE activity, greater analytic power can be attained by 
focusing on a more homogeneous population of 
studies, i.e., by studying intensively a defined set of 
PSROs, study topics, and specific criteria (or criterion 
types). The goal of the research then becomes that of 
obtaining data which are as representative as possible 
of the true population of MCE activity conducted within 
clearly defined parameters. The earliest decisions in 
the research process concerned what those parameters 
should be; the later decisions addressed the external 
validity of the findings, i.e., how results from the limited 
population thus defined could be related to the MCE 
program as a whole. 

Selection of PSROs and Time Period. The study 
focuses on PSROs which have been conducting MCEs 
for the longest time, namely the 65 PSROs which had 
achieved conditional status by mid-1975. These PSROs 
had the most opportunity to implement MCEs, since 
PSROs usually do not emphasize quality reviews until 
their utilization review programs are well established. 
By the time of the data collection period, one of the 
selected PSROs had been defunded, and two had been 
notified that their funding would not be renewed. One 
of the PSROs in the latter category elected to remain in 
the study; however, one other PSRO had reported no 
MCEs, reducing the population of PSROs eligible for 
inclusion in the sample to 62. 

A further consideration was the desirability of 
maximizing the probability that sampled MCEs would 
have completed the audit/reaudit cycle. Since reaudits 
are sometimes not performed until one year after the 
initial audit, audits performed within one year of the 
start of the anticipated data collection period 
(February 1979) were not included in the sample frame. 
The time frame thus covered initial audits completed 
between the first quarter of 1976 and February 1978. 

Selection of MCE Topics. A basic requirement of a 
variation rate study is that the MCEs assessed be 
sufficiently similar so that aggregate variation rates for 
similar criteria can be calculated. This study thus 
focuses on some of the MCE topics (diagnoses and 
procedures) which occur most frequently. Twelve of the 
twenty most frequent audit topic categories reported to 
the PSRO Management Information System (PMIS) in 
1977 were chosen based on the following additional 
criteria: 



108 



i 



• The topics represent a broad range of types of 
problems, patients, and providers. 

• Clear-cut criteria of care are available for each 
topic. 

• Topic categories are sufficiently homogeneous and 
sufficient consensus exists on elements of care. 

• Potential exists for improvements in care which 
are economically feasible. 

The following topics were chosen for study: 

Asthma 

Pneumonia (Adult) 

Congestive heart failure 

Gastroenteritis (Pediatric) 

Hypertension 

Acute myocardial infarction 

Abdominal hysterectomy 

Appendectomy 

Cesarean section (C-Section) 

Cholecystectomy 

Hip fracture 

Tonsillectomy and adenoidectomy (T&A) 

Selection and Classification of Criteria. Criteria of 
care for which variation rates were to be ascertained 
were preselected in order to minimize the burden on 
the respondents. A panel of physicians reviewed 
published criteria sets and selected three to six criteria 
in each of the twelve topic categories under study 
according to the following priorities: 

• All criteria items have a reasonable expectation of 
being included in virtually every audit on the 
specific topic. 

• Criteria are based on objective data routinely 
available in the clinical record. 

• Criteria are clearly defined with minimal need for 
interpretation on the part of the abstractors. 

• Criteria include important outcomes pertinent to 
patient health. 

• The relationship of process criteria to favorable 
outcomes is understood. 

Criteria selected by the physician panel were 
reviewed in consultation with specialists in each topic 
area. Differences were resolved in a second meeting of 
the panel. 

Criteria were further classified according to their 
relevance, as judged by the panel, to the outcomes and 



costs of care. The categories used are defined as 
follows: 

A. High efficacy criteria — criteria defining either 
outcomes that are clearly related to the 
performance of critical processes or to processes 
that are of demonstrated high efficacy in 
achieving favorable outcomes (e.g., an EKG is of 
high efficacy for diagnosing myocardial 
infarction); or are noncontroversial; or are 
appropriate for universal application in assessing 
medical care of a target population. 

B. Low efficacy criteria — criteria defining processes 
which have little or no demonstrated relationship 
to favorable outcomes or to outcomes that are of 
little long-range health benefit (e.g., "ambulatory" 
as an outcome of hospitalization for myocardial 
infarction is a low efficacy criterion); or are 
controversial; or are inappropriate for universal 
application in assessing medical care of a target 
population. 

C. Higti cost impact criteria — criteria defining 
services which have potential high impact for 
total costs of care, regardless of efficacy (e.g., 
cardiac monitoring in a coronary care unit). 

The analyses of variation rates can be performed in 
these three subgroups of criteria as well as on the 
aggregated variation rates over all the criteria. For 
instance, analysis of the change in VR from initial audit 
to reaudit can be conducted for high efficacy criteria, 
low efficacy criteria, and high cost impact criteria 
separately. The specific criteria selected and their 
classification with respect to the three subgroups 
defined above are shown in Tables 61 and 62 for 
medical and surgical topics respectively. Each 
criterion was explicitly defined in the instruction which 
accompanied the data collection form. 

The criteria in the three categories defined above 
include, for the most part, characteristics of the 
process of care, that is, diagnostic or therapeutic 
procedures. A small proportion refer to short-term 
outcomes of care, usually the patient's status at 
discharge. The three criterion types will be referred to 
as "process" criteria, to distinguish them from those 
which are more clearly outcomes of care. In addition 
to the above criteria, data on more conventional 
outcome criteria (VRs for mortality, length of stay, and 
complications) were collected for all MCEs. 



109 



TABLE 61 
Criteria for Medical Topics 



A. High Efficacy 



B. Low Efficacy 

C. High Cost Impact 



A. High Efficacy 



B. Low Efficacy 

C. High Cost Impact 

A. High Efficacy 



B. Low Efficacy 



C. High Cost Impact 



A. High Efficacy 



Asthma 

1. History of wheezing or respiratory distress unrelieved by 
usual therapy 

2. Bronchoscopy 

3. Bronchodilators 

4. Sedatives, narcotics or antihistamines 

5. Patient educated regarding disease 

6. IV fluids (hydration) 

7. Relief of symptoms by time of discharge 

1. Blood gases 

2. Chest x-ray 

3. IPPB 

4. Sputum culture 

1. Blood gases 

2. Chest x-ray 

3. IPPB 

4. Spirometry or pulmonary function tests 

5. Bronchoscopy 

Bacterial Pneumonia (Adult) 

1. Positive chest x-ray 

2. Rales and fever 

3. Elevated WBC 

4. Any appropriate antibiotics 

5. Blood and/or sputum culture 

6. Sputum smear — gram stain 

1. IPPB 

2. Discharge chest x-ray 

3. Discharge instructions 

1. Antibiotics other than penicillin or erythromycin 

2. IPPB 

Congestive Heart Failure 

1. X-ray evidence of cardiomegaly 

2. Digitalis and diuretics 

3. Daily weights 

4. Low sodium diet 

5. Patient education 

6. Serum electrolytes 

7. Physical exam demonstrating gallop and essential 
hypertension 

1. Cardiac catheterization 

2. Swan-Ganz catheter or pulmonary wedge pressure monitoring 

3. Coronary care unit 

4. History of dyspnea or orthopnea or pedal edema 

5. Serial chest x-rays 
6- EKG 

7. Complete blood count 

1 . X-ray evidence of cardiomegaly 

2. Cardiac catheterization 

3- Swan-Ganz catheter or pulmonary wedge pressure monitoring 

4. Coronary care unit 

Gastroenteristis (Pediatric) 

1. History of diarrhea and vomiting 

2. IV fluids 

3. Dehydration 



100% 

0% 
100% 

0% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 
100% 
100% 
100% 
100% 

100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 



110 



TABLE 61 (Continued) 
Criteria for Medical Topics 







4. 






5. 


B. 


Low Efficacy 


1. 
2. 
3. 


C. 


High Cost Impact 


1. 
2. 


A. 


High Efficacy 


1. 
2. 
3. 


B. 


Low Efficacy 


1. 

2. 
3. 
4. 


C. 


High Cost Impact 


1. 

2. 
3. 
4. 


A. 


High Efficacy 


1. 
2. 
3. 


B. 


Low Efficacy 


1. 
2. 
3. 
4. 


C. 


High Cost Impact 


1. 
2. 
3. 



Gastroenteritis (Pediatric) (Continued) 

Discharge feeding instructions 

Tolerating oral feedings on discharge 

Stool cultures 

Antibiotics 

Discharge Planning 

Stool cultures 

Antibiotics 

Hypertension 

Blood pressure 140/90 or higher 

Antihypertensive agents 

Patient educated regarding disease 

Laboratory tests (Calcium, electrolytes, catecholamines, BUN 

or creatinine) 

Intravenous pyelogram 

Arteriogram 

Chest x-ray 

Laboratory tests (Calcium, electrolytes, catecholamines, BUN 

or creatinine) 

Intravenous pyelogram 

Chest x-ray 

Renal arteriogram 

Myocardial Infarction 

Monitoring in CCU 

Positive EKG 

Elevated cardiac enzymes (LDH, CPK) 

Pain-free at discharge 

Ambulatory at discharge 

Chest x-ray 

Repeat or serial EKGs 

Monitoring in CCU 

Swan-Ganz catheter 

Repeat or serial EKGs 



100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 
100% 

100% 
100% 
100% 
100% 

100% 
100% 
100% 



100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



111 



TABLE 62 
Criteria for Surgical Topics 



A. High Efficacy 

B. Low Efficacy 



C. Higli Cost Impact 

A. High Efficacy 

B. Low Efficacy 

C. High Cost Impact 

A. High Efficacy 

B. Low Efficacy 

C. High Cost Impact 



A. High Efficacy 



B. Low Efficacy 



C. High Cost Impact 



A. High Efficacy 



Abdominal Hysterectomy 

1. History of dysfunctional uterine bleeding not responsive to 
therapy and/or hgb. <8 gm. 

2. Positive pathology report for fibroma or tumor 

3. Fibroma of 10 to 12 weeks size 

1. Afebrile at discharge 

2. Ambulatory at discharge 

3. Antibiotics (any) 

4. Normal bowel movement 

5. Discharge instructions given 

6. Pap smear 

1. Positive pathology report 

2. Antibiotics (any) 

Appendectomy 

1. Positive pathology report 

2. History of RLQ pain, rebound, fever or vomiting 

3. White blood count or urinalysis 

1. Chest x-ray 

2. Flat plate or plain film of abdomen 

1. Positive pathology report 

2. Chest x-ray 

3. Flat plate or plain film of abdomen 

Cesarean Section 

1. X-ray or sonographic evidence of fetal-pelvic disproportion 

2. History of previous C-section 

1. Laboratory tests (CBC or BUN or glucose or prothrombin time 
or SGOT/SGPT) 

2. Electronic fetal monitoring 

3. Instructions to patients 

4. Prophylactic antibiotics 

1. X-ray or sonographic pelvimetry 

2. Laboratory tests (CBC or BUN or glucose or prothrombin time 
or SGOT/SGPT) 

3. Electronic fetal monitoring 

Cholecystectomy 

1. Gall bladder x-rays positive for stones or 2 non-visualized 
gallbladder x-rays 

2. Positive pathology report 

3. Liver function tests (SCOT, SGPT, LDH) 

4. CBC (r/o cholangitis) 

5. History of concurrent RUQ pain 

6. Appropriate antibiotics 

1. Gl series 

2. History of fatty food intolerance or RUQ pain or jaundice 

3. Cholangiogram 

1. Gall bladder x-rays 

2. Gl series 

3. Liver function tests 

4. Cholangiogram 

Hip Fracture 

(Uncomplicated intertrochanteric fracture) 

1. X-Ray evidence of fracture 

2. Surgical fixation of fracture 



100% 

100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 
100% 

100% 
100% 
100% 
100% 
100% 

100% 



100% 

100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 
100% 



100% 
100% 



112 



TABLE 62 (Continued) 
Criteria for Surgical Topics 









Hip Fracture (Continued) 










(Uncomplicated intertrochanteric fracture) 








3. 


Stable fracture on x-ray 


100% 






4. 


Physical therapy 


100% 






5. 


Transfer to nursing home 


0% 






6. 


Pain-free at discharge 


100% 






7. 


Ambulatory at discharge 


100% 


B. 


Low Efficacy 


1. 


Afebrile at discharge 


100% 






2. 


Laboratory tests (CBC, BUN, calcium, sodium, potassium, or 
glucose) 


100% 
100% 






3. 


Follow up x-rays 








4. 


EKG 


100% 






5. 


Prophylactic antibiotics 


100% 


C. 


High Cost Impact 


1. 


Transfer to nursing home 


0% 






2. 


X-ray evidence of fracture 


100% 






3. 


Physical therapy 


100% 






4. 


Follow up x-rays 


100% 






5- 


Prophylactic antibiotics 

Tonsillectomy and Adenoidectomy 


100% 


A. 


High Efficacy 


1. 
2. 


History of bleeding tendency 

History of recurrent otitis media or airway obstruction or 

peritonsillar abscess or recurrent tonsillitis 


0% 
100% 


B. 


Low Efficacy 


1- 


Positive pathology report 








2. 


Coagulation profile 


100% 






3. 


Discharge bleeding 


100% 






4. 


Discharge temperature 


100% 


C. 


High Cost Impact 


1. 


Positive pathology report 


100% 






2. 


Coagulation profile 


100% 



5.5.2.2 SAMPLING OF MCEs 

The population of MCEs defined by the PSRO, topic, 
and time parameters established as the basis for this 
study consisted of 4,619 of the more than than 25,000 
initial audits reported to the PMIS at the time of the 
study. Simple random sampling, stratified by topic, was 
used to select 995 audits from a computer listing of the 
4,619 audits used as a sample frame. The sample size 
of 995 was judged adequate by statistical criteria and 
viable in terms of the resource constraints of the 
project and the need to minimize respondent burden. 
The distribution of studies among topics is shown in 
Table 63. 

5.5.2.3 ALLOCATING MCEs TO CRITERION TYPES 

Concerns regarding both the amount of data which 
PSROs and/or hospitals would have to collect and the 
complexity of the data collection instrument led to the 
decision that, for each MCE study, the variation rate 
from only a few criteria would be collected. Studies 
were randomly assigned to the three types of process 
criteria defined earlier (high and low efficacy, high 
cost). Since independent random samples of studies 
were drawn for each criterion type, some MCEs were 
sampled for more than one type of criterion, resulting 
in a total of 1,069 separate data requests. 



TABLE 63 
MCE Sample Count by Topic 



Topic 



Popula- e« i« Percent 

tion ^^"^P'® Sampled 



Asthma 


136 


46 


33.8 


Pneumonia (Adult) 


613 


129 


21.0 


Congestive Heart 


284 


61 


21-5 


Failure 








Gastroenteritis 


297 


64 


21.5 


(Pediatric) 








Hypertension 


242 


52 


21.5 


Myocardial Infarction 


805 


169 


21.1 


Abdominal 


383 


83 


21.7 


Hysterectomy 








Appendectomy 


493 


107 


21.7 


Cesarean Section 


373 


78 


20.9 


Cholecystectomy 


402 


82 


20.4 


Hip Fracture 


299 


63 


21.1 


Tonsillectomy and 


292 


61 


20.9 


Adenoidectomy 


4619 


995 


21.5% 



The number of studies sampled in each PSRO ranged 
from one to 77, with a mean of 15.5 studies per PSRO. The 
sample was distributed over 711 hospital sites; 493 hospitals 
had one MCE in the sample, 165 had two, 44 had three, 
and nine had more than three. 



113 



5.5.2.4 DATA COLLECTION FORMS 



5.5.2.5 DATA COLLECTION 



Process Criteria 

lEach request for data was made on a two-page form 
which specified the data elements needed, including a 
listing in random order of the criteria within the 
criterion type being sampled. The PSRO (or hospital) 
was asked to supply the variation rate for the first 
criterion, within the specified category, which was 
actually included in the original MCE audit, as well as 
the variation rate from that criterion on the reaudit. This 
allowed the examination of changes in variation rates 
by criterion type and topic, but restricted the analysis 
of individual criteria to those which appeared most 
frequently in the responses. 

Outcome Criteria 

In additibh to the process criteria previously defined, 
variation rates were requested for the following 
outcome criteria: 

• length of stay (any length of stay criterion used) 

• mortality 

• complications (specific list for each topic) 

Unlike the process criteria, these were requested for 
each audit and reaudit sampled. 

Hospital Information 

Data were collected for each hospital which served 
as an MCE study site concerning its size, teaching 
status, rural/urban location, ownership, and MCE 
delegation status. 

Reaudit Data 

The same data were requested for reaudits as for 
audits, with two qualifications. First, the number of 
variations judged after peer review to be unjustified 
was requested for criteria at audit, but not at reaudit. 
This decision was based on the likelihood that many 
reaudits would have to be performed and that the need 
for physician committees to review reaudit results 
would seriously impair response rates. Second, the 
sample size requested for reaudits was held to the 
minimum judged necessary for analysis, i.e., ten or 
more patient records. Although this is considerably 
smaller than the sample sizes commonly found in 
reaudits, the low sample size was deemed desirable 
since most data requests would require new reaudit 
data to be collected. This was expected both because 
reaudit rates in general have been low, and because 
the design of the study required reaudit data even for 
criteria which showed no deficiencies at the initial 
audit. Of prime concern in asking PSROs to perform 
reaudits was that the criteria reaudited be identical to 
those audited; methods of sampling patient records 
were left to the PSRO and/or hospital, except that the 
reaudit samples involve patients discharged at least 
three months after the initial audits were completed. 



The success of this study depends on the voluntary 
cooperation of the sampled PSROs and hospitals. 
Consequently, every effort was made to minimize the 
burden on the respondents, to make the data requests 
clear and simple, to provide assistance, and to 
persuade PSROs and through them their hospitals of 
the importance of the study and of their participation 
in it. 

PSRO Executive Directors and their MCE staffs were 
initially informed of the existence and nature of the 
study in letters from HSQB in November 1978. Data 
collection packages were sent to the PSROs in late 
May and early June of 1979. These consisted of 
explanatory letters and, for each sampled MCE, a 
two-page data collection form with instructions and 
criteria definitions. The forms, instructions, and 
definitions had been refined on the basis of pretests 
in three PSROs. Each PSRO decided how best to 
collect the information, whether by its own personnel 
or by review coordinators in the hospitals. Data 
elements which were known in advance, such as 
characteristics of the sampled MCEs obtained from 
HSQB 131 Forms, were preceded on the data forms so 
that respondents could verify them; sampled criteria of 
care and complications for which variation rates were 
requested were also entered on each form. 
Confidentiality was promised in that hospitals were not 
identifiable, and neither hospitals nor PSROs would be 
identified in the analysis. Telephone contacts were 
maintained with most of the study PSROs through all 
stages of the data collection. 

5.5.2.6 CHARACTERISTICS OF RESPONSES 

A review of the relationship between the sample 
requested and the data actually obtained was made to 
assure that the overall response rate was sufficient for 
analysis and that the data were unbiased with regard 
to the characteristics of the respondents. Table 64 
shows the overall response rate and the net response 



TABLE 64 
Response Rates 





Number 


Percent 


Data forms sent out 


1069 


100% 


Data forms returned 


889 


83 


Data forms returned but 


121 


11 


excluded: 






Topic not included in study 


30 


3 


Not filled out 


52 


5 


Areawide audits; data on 


10 


1 


individual hospital not 






available 






Uncorrectable inaccuracies 


19 


2 


or omissions 






Net responses used in analysis 


768 


72% 



114 



rate after forms which were not usable were excluded. 
Due to the overlapping sampling methods described 
above, the 768 forms returned came from 710 separate 
MCEs. 

The study required a good deal of work on the part of 
the sampled PSROs; for many of the larger PSROs with 
numerous MCEs in the sample, this amounted to a 
considerable burden at a time when resources were 
strained. It is on their willing and often enthusiastic 
participation that this study rests. Only six of the 62 
PSROs in the sample chose not to participate, 
eliminating 92 (8.6%) of the possible responses. Of the 
56 participating PSROs, 36 had a usable response rate 
of over 75 percent of the sampled MCEs, and only 10 
had a usable response rate under 50 percent. 

Responses were not noticeably patterned over the 12 
topics or the 10 HEW regions. The usable response 
rate by topic varied from 56 to 85 percent, while the 
regional response rates ranged from 53 to 99 percent. 

Within the usable responses, variation rates for the 
requested criteria were available approximately 90 
percent of the time. The exceptions occurred primarily 
when the criteria requested were not used in the 
sample audits. There was some variation by the type of 
criterion, as follows: 



Criterion Type 



Percent availability 

among usable 

studies 



Process 

High efficacy 
Low efficacy 
High cost 

Outcome 

Length of stay 

Mortality 

Complications 



90 
74 
67 

92 
94 
88 



The differences in availability among the three types 
of process criteria (which were sampled equally) can 
be interpreted favorably: high efficacy criteria (those 
most closely related to favorable patient outcomes) are 
more likely to be found on MCE criteria lists than those 
which have less relation to outcomes (low efficacy) or 
those whose primary relationship is to the cost of 
care (high cost). 

An examination of the dates when the initial audits of 
the sample MCEs were completed reveals that less 
than 1 percent were done in 1975, 32 percent in 1976, 
59 percent in 1977, and 8 percent in 1978. The sample 
is thus concentrated among studies begun in 1976 and 
1977, and does not reflect possible later changes due 
to the effects of individual PSROs' learning curves or 
changes in policy at the national level. On the other 
hand, the PSROs under study are the earliest cohort to 
achieve conditional status and presumably had 
implemented MCEs fully in this period. Case studies to 
be performed in six PSROs will describe subsequent 
changes in MCE programs. 



Reaudit dates for the sample MCEs are considerably 
more recent. Reaudits had been done for 31 percent of 
the final sample, but often they did not include the 
requested criteria in exactly the same form as the initial 
audit, necessitating a second reaudit on identical 
criteria. Ultimately, 82 percent of reaudits were done 
especially for this study. Further analyses will assess 
the effect of the intervening time between audit and 
reaudit on measured changes in variation rates. 

5.5.3 Methods of Analysis 

5.5.3.1 BASIC CONCEPTS 

When planning an MCE audit or reaudit on a given 
topic, the typical medical audit committee decides, 
among other things, the explicit criteria which specify 
important dimensions (or characteristics) of 
compliance. Each criterion is accompanied by a 
standard, generally a statement that all patient records 
should have the characteristic (100% for desirable 
characteristics) or that none should (0% for 
undesirable characteristics). Allowable exceptions to 
the standard for each criterion are also specified in 
advance. The sampled patient records are then 
examined according to the criteria, and the proportion 
complying or not complying with the standard for each 
criterion is recorded. These tallies then serve as the 
basis for the committee's analysis and findings. 

In examining the deviations from standards revealed 
by the tallies, the audit committee may wish to exempt 
certain of them as justified, and take action only on 
the remaining ones. The process of justification must 
be carefully distinguished from the specification of 
allowable exceptions to criteria before the data are 
gathered: the exception process is explicit; the 
justification process is implicit and, from an evaluation 
standpoint, subject to unreliable application. For this 
reason as well as the lack of systematic data on the 
justification process, the primary statistical analyses of 
variation rates in this study will be based on deviations 
from standards regardless of any jurisdictions 
subsequently invoked when medical audit committees 
analyze the data. This is not to deny the need for 
clinical judgment in quality assessment or the 
desirability of examining variations which have been 
subjected to physician review, but rather to assert that, 
for some purposes at least, the raw variation rate can 
be a valuable index of change in the quality of care. 

The variation rates to be compared, then, are the 
proportions of patients audited whose care deviated 
from the standard established for a particular criterion 
before (i.e., at initial audit) and after (at re-audit) 
corrective actions presumably were taken in the MCE 
cycle — regardless of whether the deviations were 
regarded as justified. Audit and reaudit variation rates 
are derived from different samples of patient records at 
each study site. This is basic to the MCE process: the 
samples at audit and re-audit represent successive 
cross-sections of medical care before and after the 



115 



ntervention. The variation rates at audit (VRJ and at 
/eaudit (VR2) are thus paired. 

Except for a limited number of specific criteria which 
have sufficiently large sample sizes, the comparisons to 
be made will be of types of criteria. The most important 
types, identified earlier, areas follows: 

Process (primarily specifying diagnostic and 
therapeutic procedures, but including 
some outcomes) 

High efficacy 

Low efficacy 

High cost 
Outcome (end results of care) 

Length of stay 

Mortality 

Complications 

Variation rates for these criteria types will be compared 
both within and across the 12 topics studied. 

5.5.3.2 TESTING FOR CHANGE 

The analysis begins with the basic question: Do the 
data indicate that improvement in compliance to 
standards of care and outcome (reduction in variation 
rates) occurred between the time of audit and reaudit? 
The null hypothesis is that there is no difference in the 
audit and reaudit VR distributions. 

Two tests of significance are used, with somewhat 
different implications. The first, the Wilcoxon signed 
ranks test, is a nonparametric test which looks at the 
sign and magnitude of VR differences for each audit- 
reaudit pair. If the test statistic produced is significant 
(at the .10 level), the medians of the audit and reaudit 
VR distributions are judged to be different. The 
Wilcoxon test, by weighing equally all sites at which a 
particular category of criterion was used, addresses 
the question of whether hospitals using MCEs as a 
method of quality assurance tended to improve their 
care or not. 

A second procedure, the Z test, is used to test 
whether care improved over all patients included in 
MCEs. This parametric test simply compares the 
pooled samples at audit and reaudit and their variation 
rates, regardless of the VR findings at each site. The 
Z-test is thus less sensitive to sample size limitations. 
It is likely, therefore, that more significant differences 
will be found using this method since the sample sizes 
available by combining all the studies in a specified 
topic and criterion type can be very large and hence 
smaller differences are more likely to be found 
significant. The level of significance chosen as a 
threshold for the Z test is again .10. 

Application of the Wilcoxon and Z tests to various 
categories of criteria or topics yields multiple measures 
of significance. These are in turn summarized by an 
overall test of the (binomial) probability that a given 
number of tests would be found significant purely 
through chance. 

After establishing overall changes, a subset of 
criteria which experienced initial variation rates of 10 



percent or more was examined in the same way. The 
rationale for excluding initial variation rates under 10 
percent is that (a) they provide less room for 
improvement and (b) for the most part they would not 
be the focus of an MCE committee's attention. The 
issue for this part of the analysis is whether the areas 
in which MCEs detect larger problems in care, and 
which are more likely to be acted upon, actually 
improve between audit and reaudit. 

5.5.4 Limitations 

A number of factors must be considered in 
intrepreting the results of this study. Some are inherent 
in the study design and methods of data collection, 
while others need to be taken into account only when 
considering certain uses of the data. They are 
presented here together with strategies which may be 
used to mitigate their influence. 

Response Bias. Whenever the response rate is less 
than 100 percent, the possibility exists that the actual 
responses are a biased sampling of those intended. 
The primary alternative to the null hypothesis is that the 
results differ due to self-selection of respondents. In 
this study, the overall response rate is 72 percent, a 
sizeable proportion for a study which requires a great 
deal of effort on the part of the respondents. One may 
be fairly comfortable that self-selection of cooperating 
PSROs is not a major factor: as discussed above, only 
six of 62 PSROs chose not to participate. At the 
hospital level, some self-selection may have taken 
place; there is no way of knowing to what degree the 
responding hospitals represent the intended sample. 

Biased Reporting. A more serious challenge to 
validity is the possibility that the data used to calculate 
variation rates are not accurately reported, but include 
a bias commonly found in social surveys: respondents 
tend to want to make themselves "look good." This 
may occur either in the data collected for the MCE 
themselves or in the abstractions done specifically for 
this study. It is expected that this bias occurred less in 
the data collection for this study than in most surveys, 
since the data were reported by medical records 
professionals. 

Overrepresentation of Larger PSROs and IHospitals. 
The sample for this study is based upon MCEs which 
have been reported to HSQB. One consequence of this 
fact is that larger PSROs and larger hospitals are 
heavily represented, especially since the numbers of 
MCEs required by the PSRO program have been 
proportional to hospital size. The aggregate findings 
of the study will thus depend more heavily on what is 
going on in these sites in terms of changes in care. In a 
sense, this properly represents the results of the 
population of MCEs, but for some purposes the analysis 
must be stratified by PSRO or hospital size. 

Attribution. This is perhaps the most serious 
challenge to the study. Measurements of changes in 
compliance to criteria of care can only determine if 



116 



something happened between the audits and reaudits, 
but cannot specify what caused the changes. Actions 
taken as a result of the initial audit were not collected. 
In addition, it is difficult to know what interventions 
other than MCE, such as changes in technology or 
prevailing practices, may be responsible. Ideally, 
a set of randomized control data from non-MCE sites 
could provide answers; however, this is not currently 
feasible. The inference that changes in variation rates 
can be caused by MCEs can, however, be strengthened 
by examining some MCEs in detail to trace the process 
of change. This is being done now in a set of case 
studies which will be available in the spring of 1980. 

Small Samples at Reaudit. In order to minimize the 
burden on the respondents, they were asked to reaudft 
a smaller number of patient records (10 or more) than 
is normally accepted in audit methodology. The 
proportions at variation in the reaudit are thus 
sometimes based on small denominators in each MCE. 
This is less of a problem for aggregate analysis, 
however, than it would be for an individual MCE. 

Regression to the Mean. A portion of the analysis 
considers what happens at reaudit to criteria which 
had a variation rate at audit greater than or equal to an 
arbitrary cutting point, 10 percent. The primary 
rationale for this is simple: these criteria are more 
likely to reflect significant problems in care. Yet, this 
tactic invites a familiar artifact of measurement, that 
high initial values are likely to become lower and low 
values higher at the second measurement. This 
phenomenon ("regression toward the mean") applies 
regardless if the measure is of IQ, knowledge, or 
variation rate. 

Difficulties of Extrapolation. The results of this study 
are, strictly speaking, applicable only to the MCEs and 
criteria sampled. The issue is whether the MCE proc- 
ess, as a tool, is related to changes in physician be- 
havior and patient outcome in these sites and for these 
dimensions of care. Attempts to extend the results to 
the MCE component of the PSRO program as a whole 
require assumptions which are often tenuous. 

Effect on Federal Beneficiaries. Each MCE is a 
study of the care rendered to a sample of all patients 
in the specified topic area. At no point in the MCE 
process or in this evaluation are Federal patients dis- 
tinguished from non-Federal patients. The PSRO pro- 
gram assumes that improvements in quality of care 
will affect all patients alike; this assumption cannot be 
tested with the available data. 

Global Effects. A limitation intrinsic to the study 
design is the inability to measure global effects of the 
program. If the MCE program improved quality of care 
simply by making physicians and others aware that 
their performance would be scrutinized, then the pro- 
gram effect would precede the initial audit. This im- 
pact would not be detected through variation rate 
analysis unless appropriate controls were included in 
a quasi-experimental design. There is reason to be- 
lieve, on the other hand, that such global effects are 



not large or sustained, and that the program would 
have to show concurrent improvements in quality in 
order to be considered effective. 

Time Between Audit and Reaudit. The audits sam- 
pled took place primarily in 1976 and 1977, while 
most of the reaudits were done in 1979. The PSRO 
program recommends that reaudits be done within a 
year of the initial audit. This study thus measures 
longer-range changes than are usually tapped by re- 
audits. While these changes are more enduring, more 
has happened in the environment that may have caused 
them, or attenuated them. 

Early MCE Activity. Many PSROs have asserted that 
their more recent MCE activity is more effective than 
that examined in this study. The study PSROs were, 
however, among the earliest to achieve conditional 
status and were well established in 1976 and 1977. The 
case studies to appear in the spring will help to bal- 
ance the picture by examining changes in the MCE 
programs of six PSROs between that period and the 
present. 

5.5.5 Findings 

5.5.5.1 TESTS OF SIGNIFICANCE FOR ALL 
STUDIES 

The first question to be addressed is whether the 
data show favorable changes in care. Table 65 pre- 
sents a summary of variation rate changes for the six 
types of criteria studied, using both the Wilcoxon and 
Z tests of significance. Data from all audits and re- 
audits were used to compute the Z tests; only paired 
data were used in the Wilcoxon tests. The Wilcoxon 
thus compares criteria which are exactly identical from 
audit to reaudit, within each criterion type, while the 
Z test includes a small fraction of VRs (less than 3%) 
which are unpaired. 

High efficacy, high cost, and mortality criteria ex- 
hibited the most clearly significant improvements, while 
for the low efficacy and complications categories only 
the two-sample Z test showed significant results. The 
latter finding is most likely attributable to the fact that 
the large pooled samples used in the Z test make 
smaller differences more significant. No change was 
found in length of stay using either method. 

The finding of a strong overall effect among high 
efficacy criteria is encouraging. High efficacy criteria 
are those judged at the outset to be most clearly 
related to adequate care process and good patient 
outcome. Low efficacy criteria, by contrast, showed 
borderline changes at best. These are the criteria 
deemed least likely to distinguish good from bad care. 
Improvement in high cost criteria indicates that costly 
elements of care are used more frequently at reaudit 
than at audit. 

Length of stay was affected least of all the 
criterion types. The length of stay category is the 
most difficult of the six to interpret, since it contains 
the most heterogeneous criteria. The data analyzed 



117 



TABLE 65 
Summary of Variation Rate Changes — All Topics — All Criteria 



Initial Audit 



Reaudit 



Criterion 
Type 



Number 

of 
Audits 



Number 

of 
Patients 



Proportion 

at 

Variation 

(VRJ 



Number 

of 
Reaudits 



Number 

of 
Patients 



Proportion 

at 

Variation 

(VR.) 



P Value 



High Efficacy 


233 


8446 


.154 
.153 


233 


4202 


.136 
.123 


.05 
<.005 


Low Efficacy 


187 


6837 


.125 
.112 


186 


3138 


.118 
.105 


n.s.* 
.07 


High Cost 


172 


6573 


.145 
.125 


173 


2997 


.109 
.108 


.01 
.08 


Length of Stay 


650 


24007 


.196 
.182 


637 


10759 


.204 
.189 


n.s. 
n.s. 


Mortality 


670 


25067 


.062 
.054 


672 


11876 


.056 
.050 


.01 
.03 


Complications 


622 


112396** 


.044 
.046 


606 


52771 


.045 
.042 


n.s. 
<.005 



NOTE: The first of the two proportions in each VR cell is the mean of the VR of each audit or reaudit, and is related to 
the Wilcoxon test. The second proportion is the pooled VR for all patients to whom the criteria were applied, 
over all the audits or reaudits. The Z-test evaluates the differences in these proportions. In the probability ("P 
value") column, the first probability in each cell was computed by the Wilcoxon test, the second by the Z-test. 
These probabilities denote the likelihood that the difference between VRi and VR;. is attributable solely to chance. 

* not significant — probability greater than 0.10. 

** The denominator for complication VRs is large because multiple complications were studied in most MCEs. 



are not actual lengths of stay, but variations from 
standards set by each MCE committee. Not only may 
each MCE committee choose different thresholds for 
maximum length of stay, but it may examine pre- 
operative, post-operative, and short stays as well. 

The difference in mortality rates is significant by 
both tests, although there is no apparent change in 
the overall proportion of deaths when reported to two 
decimal places. This indicates that the changes which 
took place were small but consistently favorable over 
the sample of MCEs and patients. 

Complication rate changes were not significant by 
the paired Wilcoxon test but highly significant by 
the two sample Z-test. Apparently, there are move- 
ments in both directions in individual MCEs, but the 
MCEs with favorable differences had more patients 
audited and/or larger changes so that the Z test 
proved significant in the favorable direction. 

Table 66 shows the areas in which significant 
changes were found, by topic and type of criterion. 
A "W" or "Z" in a cell of this table indicates which 
specific test showed a significant change in VR at 
the .10 level. Parentheses around a "W" or "Z" 
indicate that the change was in the unfavorable 
direction: that is, the variation rates were observed to 



increase from audit to reaudit. 

The cells with both "W" and "Z" indicate improve- 
ments in care, as measured by significant decreases 
in VR using both tests. These cells include: 

• High efficacy criteria in pneumonia, asthma, and 
T&A. 

• Low efficacy criteria in pneumonia and 
cholecystectomy. 

• High cost criteria in appendectomy, abdominal 
hysterectomy, and Cesarean section- 

• Mortality in congestive heart failure and asthma. 

• Complications in Cesarean section. 

Cells with significant unfavorable changes for both 
tests as indicated by both (W) and (Z), include: 

• High efficacy criteria in appendectomy and 
Cesarean section. 

• Length of stay in appendectomy. 

• Mortality in hypertension. 

The next two tables, 67 and 68, apply significance 
tests to the number of comparisons found to be 
significant. The question here is: "How likely is it 
that the results of the Wilcoxon or Z-tests could 
have occurred by chance?" Table 67 examines this 
issue by criterion type and Table 68 by topic. For 



118 



TABLE 66 
All Studies, Significant Results by Criterion Type and Topic 



Criterion Type 



High 
Efficacy 



Low 
Efficacy 



High 
Cost 



LOS 



IVIortality 



Complica- 
tions 



Topic: 
Gastroenteritis 








(Z) 




(Z) 






Hypertension 




z 


w 






z 


(W) 




AMI 








(Z) 


w 




w 




CHF 












(Z) 






Pneumonia 


W 


z 


w 


z 




z 




z 


Asthma 


w 


z 




(Z) 




(Z) 






Hip Fracture 




z 












(Z) 


T&A 


w 


z 












(Z) 


Appendectomy 


(W) 


(Z) 




(Z) 


w 


z 


(W) 


(Z) 


Cholecystectomy 


w 




w 


z 








(Z) 


Ab. Hysterectomy 




(Z) 




z 


w 


z 




(Z) 


C-Section 


(W) 


(Z) 






w 


z 




z 



(W) (Z) 



w z 



w z 



(W) 



w 



(Z) 



Criteria: 



Total 
Possi- 
ble 
Tests 





WilcoxonTest 






Z Test of Pre 


Num- 


Num- 




Num- 


Num- 


ber of 


ber of 




ber of 


ber of 


Posi- 


Nega- 




Posi- 


Nega- 


tive 


tive 




tive 


tive 


Find- 


Find- 




Find- 


Find- 


ings 


P(+) ings P(-) 


NS 


ings 


P(+) ings 



P(-) 



(Z) 

z 

(Z) 



w 
w z 



KEY: W — Wilcoxon test significant at the .10 level. 

Z — 2-sample test of proportion significant at the .10 level. 

Parentheses indicate significant increases in variation rate from audit to reaudit. 

TABLE 67 
Summary of Significant Comparisons by Criterion Type — All Studies 



NS 



High Efficacy 
Low Efficacy 
High Cost 


12 
12 

12 


4 
3 
4 


.026 
.111 
.026 


2 




.341 


6 
9 
8 


5 
3 
5 


.0043 

.111 

.0043 


3 
4 
3 


.111 
.026 
.111 


4 
5 
4 


Sum of 
Above 3 


36 


11 


.0001 


2 


.19 


23 


13 


.0001 


10 


.0002 


13 


LOS 

Mortality 

Complications 


12 
12 
12 


1 
2 
3 


.718 
.341 
.111 


2 

1 

1 


.341 
.718 
.718 


9 
9 
8 


2 
2 

4 


.341 
.341 
.026 


5 
2 
2 


.0043 

.341 

.341 


5 
8 
6 


Sum of 
Above 3 


36 


6 


.075 


4 


.41 


26 


8 


.0073 


9 


.001 


19 


Total of All 6 


72 


17 


.0001 


6 


>.5 


49 


21 


.0001 


19 


.0001 


32 



"Positive findings" indicate statistically significant reductions in VR from audit to reaudit. 
"Negative findings" indicate statistically significant increases in VR from audit to reaudit. 
"NS" indicates the number of findings which were not significant. 

"P(+)" and "P(— )" denote the binomial probability that the indicated number of positive (or negative) findings could have 
occurred solely by change. Values with P(+) and P(— ) less than .10 are considered significant. 



119 



TABLE 68 
Summary of Significant Comparisons by Topic — All Studies 





Total 




WilcoxonTest 






Z Test of Proportions 






Num- 




Num- 






Num- 




Num- 








Pn^€ii- 


ber of 




ber of 






ber of 




ber of 






Topic 


ble 
Tests 


Posi- 
tive 
Find- 




Nega- 
tive 
Find- 






Posi- 
tive 
Find- 




Nega- 
tive 
Find- 










ings 


P(+) 


ings 


(P-) 


NS 


ings 


P(+) 


ings 


(P-) 


NS 


Gastroenteritis 


6 














6 


1 


.469 


2 


.114 


3 


Hypertension 


6 


1 


.469 


2 


.114 


3 


2 


.114 


1 


.469 


3 


AlVII 


6 


2 


.114 





— 


4 


1 


.469 


1 


.469 


4 


CHF 


6 


1 


.469 


1 


.469 


4 


1 


.469 


1 


.469 


4 


Pneumonia 


6 


2 


.114 





— 


4 


4 


.0013 


1 


.469 


1 


Astiima 


6 


2 


.114 





— 


4 


3 


.016 


2 


.469 


1 


Sum: Medical 


36 


8 


.0073 


3 


>.5 


25 


12 


.0001 


8 


.0073 


16 


Hip Fracture 


6 


1 


.469 








5 


1 


.469 


1 


.469 


4 


T&A 


6 


1 


.469 





— 


5 


1 


.469 


3 


.016 


2 


Appendectomy 


6 


1 


.469 


2 


.114 


3 


1 


.469 


3 


.016 


2 


Cholecystec- 
























tomy 


6 


2 


.114 





— 


4 


1 


.469 


1 


.469 


4 


AB. Hysterec- 
























tomy 


6 


2 


.114 





— 


4 


2 


.114 


2 


.114 


2 


C-Section 


6 


2 


.114 


1 


.469 


3 


3 


.016 


1 


.469 


2 


Sum: Surgical 


36 


9 


.0013 


3 


>.5 


24 


9 


.0013 


11 


.0001 


16 


Sum: All Topics 


72 


17 


.0001 


6 


>.5 


49 


21 


.0001 


19 


.0001 


32 



"Positive findings" indicate statistically significant reductions in VR from audit to reaudit. 
"Negative findings" indicate statistically significant increases in VR from audit to reaudit. 
"NS" indicates the number of findings which were not significant. 

"P(+)" and "P(— )" denote the binomial probability that the indicated number of positive (or negative) findings could have 
occurred solely by chance. Values with P(+) and P(— ) less than .10 are considered significant. 



example, in Table 68, the probability of finding four 
of twelve possible Wilcoxon tests (one for each topic) 
for high efficacy criteria to be significant at the .10 
level if there were no underlying relationships was 
found to be small, i.e., .026. The positive findings 
for high efficacy and high cost criteria are seen to be 
significant by both tests, while the positive findings 
for complications are significant only by the Z-test. 

The negative changes detected by the Wilcoxon 
test in Table 67 were not significant, but the Z-test 
showed significant negative changes by low efficacy 
criteria and length of stay. 

None of the topics in Table 68 show significant 
positive findings by the Wilcoxon test, although six 
approach the .10 level. This may be a consequence 
of the small number of MCEs sampled for each topic, 
since the subtotals for medical and surgical topics as a 
group, as well as the total for all topics combined, 
show significant numbers of positive findings- The 
negative findings, based on the Wilcoxon test, are 
not significant. 

The Z test, as expected, shows more topics with 
significant numbers of findings, three positive 
(pneumonia, asthma, and Cesarean section) and two 
negative (T&A and appendectomy). The Z test, which 
indicates what changes are taking place in the 



pooled patient samples, thus shows movement in 
both directions with no distinct positive or negative 
trend. 

5.5.5.2 TESTS OF SIGNIFICANCE FOR CRITERIA 
WITH INITIAL VR ^ .10 

Examination of only those criteria with initial 
variation rates at or above an arbitrary threshold of 
ten percent shows significant improvements for all 
criterion types (Table 69). This tends to confirm the 
hypothesis that those criteria which are more likely 
to be acted upon, and to represent serious problems 
in care, exhibit more improvement. Table 70 provides 
further confirmation by showing a large number of 
positive significant differences for various combinations 
of topics and criterion types and only four negative 
ones. The probabilities that these numbers of 
significant improvements in care and outcome would 
occur by chance alone are very small, as shown in 
Table 71 by type of criterion and in Table 72 by topic. 
All cases of negative findings are seen to have chance 
probabilities, i.e., they are nonsignificant. 

The findings in Table 71 are uniformly positive: 
the numbers of positive findings for all criterion types 
are statistically significant by both tests, and negative 



120 



Criterion 
Type 



TABLE 69 

Summary of Variation Rate Changes — All Types — Criteria with Initial Variation 
Rates 10 Percent or More 



Initial Audit 



Reaudit 



P Value 



Number 

of 
Audits 



Number 

of 
Patients 



Proportion 

at 

Variation 

(VRx) 



Number 

of 
Reaudits 



Number 

of 
Patients 



Proportion 

at 

Variation 

(VR.) 



High Efficacy 


89 


3445 


.376 
.349 


90 


2007 


.203 
.166 


<.005 
<.005 


Low Efficacy 


56 


2043 


.376 
.332 


'56 


946 


.235 
.212 


<.005 
<.005 


High Cost 


66 


2414 


.345 
.300 


67 


1087 


.195 
.209 


<.005 
<.005 


Length of Stay 


373 


13489 


.323 
.302 


372 


6211 


.260 
.244 


<.005 
<.005 


Mortality 


165 


5876 


.219 
.196 


184 


2989 


.162 
.149 


<.005 
<.005 


Complications 


77 


14943 


.181 
.169 


79 


7012 


.099 
.096 


<.005 
<.005 



NOTE: The first of the two proportions in each VR cell is the mean of the VR of each audit or reaudit, and is related to 
the Wllcoxon test. The second proportion Is the pooled VR for all patients to whom the criteria were applied, over 
all the audits and reaudits. The Z-test evaluates the differences In these proportions. In the probability ("P value") 
column, the first probability In each cell was computed by the Wllcoxon test, the second by the Z-test. These 
probabilities denote the likelihood that the difference between VRi and VR2 Is attributable soley to chance. 



findings are absent or nonsignificant. When examined 
by topic (Table 72), three medical topics and one 
surgical topic show statistically significant numbers 
of positive Wllcoxon tests; a number of other topics 
are close (P=.114). All but one topic (gastroenteritis) 
show significant numbers of positive Z tests. These 
findings provide persuasive evidence that MCEs are 
improving care for those criteria which had substantial 
variation to begin with. 

Regression toward the mean undoubtedly has had 
some influence on these findings, making the MCE 
effect appear greater than it actually is. However, 
there is reason to believe that this regression effect is 
not great, and that a genuine MCE effect is shown- 
For one thing, the threshold used (VRi^.10) is not 
extreme; in fact, it is less than the mean of the initial 
variation rates shown in Table 65. Indeed a counter- 
argument could be made that those initial variation 
rates below .10 would have an artifactual tendency to 
■rise to the mean and thereby dilute the overall estimate 
of MCE effects. This is especially true of those 
criteria sampled for this study which had no variations 
at initial audit: they could exhibit no changes other 
than increases in VR. 



5.6 Benefits and Costs of Medical Care 
Evaluation Studies 

5.6.1 Objectives 

This study represents a first attempt to assess the 
health benefits of MCEs in relation to their costs 
and the costs of changes in care which they bring 
about. This is done in two steps: first, MCE health 
benefits are estimated and translated into dollar values; 
second, costs of MCEs and of changes in care are 
calculated and compared to the dollar value of MCE 
health benefits to produce a benefit-cost estimate. 
No attempt is made to generalize study results beyond 
the MCEs and PSRO hospitals which provided data 
for the study. 

5.6.2. Methods 

The overall strategy of the study is summarized 
in Figure 23. Variation rates and variation rate changes 
were collected for 105 MCEs done in eight 
experienced PSROs in four MCE topic areas. The 
topics (congestive heart failure, asthma, cholecyst- 



121 



TABLE 70 

Criteria with Initial Variation Rates of 10 Percent or More: Significant Results 
by Criterion Type and Topic 



Criterion Type: 


High 
Efficacy 


Low 
Efficacy 


High 
Cost 


LOS 


Mortality 


Complica- 
tions 


Topic: 
























Gastroenteritis 
















z 






W Z 


Hypertension 


W 


z 






W 


z 










Z 


AMI 


W 


z 


W 


Z 


W 


z 


w 


z 


W 




w z 


CHF 




z 












z 


W 


z 


z 


Pneumonia 


W 


z 


W 


z 






w 


z 


W 


z 


w z 


Asthma 


W 


z 




(Z) 




(Z) 




z 






w z 


Hip Fracture 


W 


z 


W 


z 






w 




W 


z 


w z 


T&A 


W 


z 








z 




z 








Appendectomy 




(Z) 




(Z) 


w 


z 




z 






w z 


Cholecystectomy 




z 




z 






w 






z 


z 


AB. Hysterectomy 






W 


z 




z 




z 






w z 


C-Section 










w 


z 


w 


z 






z 



KEY: W — Wilcoxon test significant at the .10 level. 

Z — 2-sample test of proportion significant at the .10 level. 

Parentheses indicate significant increases in variation rate from audit to reaudit. 



TABLE 71 

Summary of Significant Comparisons by Topic-Criteria with Initial Variation 
Rates of 10 Percent or More 



Criteria: 



Total 
Possi- 
ble 
Tests 





Wilcoxon Test 








Z Test of Pro| 


Num- 


Num- 






Num- 


Num- 


ber of 


ber of 






ber of 


ber of 


Posi- 


Nega- 






Posi- 


Nega- 


tive 


tive 






tive 


tive 


Find- 


Find- 






Find- 


Find- 


ings 


P(+) ings 


P(-) 


NS 


ings 


P(+) ings 



P(-) 



NS 



High Efficacy 
Low Efficacy 
High Cost 


12 
12 
12 


6 
4 
4 


.0005 

.026 

.026 







— 


6 
8 

8 


8 
5 
5 


.0001 
.0043 
.0005 


1 
2 

1 


>.5 

.341 
>.5 


3 
5 
5 


Sum of 
Above 3 


36 


14 


.0001 





— 


22 


19 


.0001 


4 


.413 


13 


LOS 

Mortality 
Complications 


12 
12 
12 


5 
4 

7 


.0043 

.026 

.0001 







— 


7 
8 
5 


9 

4 

11 


.0001 

.026 

.0001 







— 


3 
8 

1 


Sum of 
Above 3 


36 


16 


.0001 





— 


20 


24 


.0001 





— 


12 


Total of All 6 


72 


30 


.0001 





— 


42 


43 


.0001 


4 


>.5 


25 



"Positive findings" indicate statistically significant reductions in VR from audit to reaudit. 
"Negative findings" indicate statistically significant increases in VR from audit to reaudit. 
"NS" indicates the number of findings which were not significant. 

"P(-f)" and "P(— )" denote the binomial probability that the indicated number of positive (or negative) findings could have 
occurred solely by chance. Values with P(+) and P(— ) less than .10 are considered significant. 



122 



TABLE 72 



Summary of Significant Comparisons by Topic-Criteria with Initial Variation 
Rates of 10 Percent or More 



Wilcoxon Test 



Z Test of Proportions 



Topic 



Total 
Possi- 
ble 
Tests 



Num- 
ber of 
Posi- 
tive 
Find- 
ings 



P(+) 



Num- 
ber of 
Nega- 
tive 
Find- 
ings 



P(-) 



NS 



Num- 
ber of 
Posi- 
tive 
Find- 
ings 



P(+) 



Num- 
ber of 
Nega- 
tive 
Find- 
ings 



P(-) 



NS 



Gastroenteritis 

Hypertension 

AMI 

CHF 

Pneumonia 

Asthma 



.469 

.114 

.0001 

.069 

.0001 

.114 












5 
4 

5 
1 
4 



.114 

.016 

.0001 

.0013 

.0001 

.016 



.114 



Sum: Medical 



36 



17 



.0001 



— 19 



22 



.0001 



>.5 



12 



Hip Fracture 

T&A 

Appendectomy 

Cholecystec- 
tomy 

AB. Hysterec- 
tomy 

C-Section 



.0001 

.469 

.114 

.469 

.114 
.114 



— 1 

— 5 

— 4 

— 5 

— 4 

— 4 



.0013 

.016 

.016 

.0013 

.0013 
.016 



.114 



"Positive findings" indicate statistically significant reductions in VR from audit to reaudit. 
"Negative findings" indicate statistically significant increases in VR from audit to reaudit. 
"NS" indicates the number of findings which were not significant. 

"P(+)" and "P(— )" denote the binomial probability that the indicated number of positive (or negative) 
occurred solely by chance. Values with P(+) and P(— ) less than .10 are considered significant. 



Sum: Surgical 


36 


13 


.0001 





— 


23 


21 


.0001 


2 


>.5 


13 


Sum: All Topics 


72 


30 


.0001 





— 


42 


43 


.0001 


4 


>.5 


25 



findings could have 



Figure 23 
Overview of Benefit-Cost Strategy 



DATA 



INTERPRETATION 



ESTIMATES 



.Literature and 
data reviews 

.Variation rates 
for all criteria 
in studies for 
A topics, 8 PSROs 



-j>- 



Physlcian 
Panel 



•Health status 
impact of MCEs 



.Costs of MCEs 
.Hospital charge 
data 
.Changes in care 



Economic 
Analysis 



.Changes in cost 
incident to MCEs 

.Dollar value of 
health benefits 



Benefit-Cost 
" Analysis 



123 



ectomy, and Cesarean section) were chosen 
because they constitute frequent causes of hospital- 
ization, are frequent subjects of MCEs, and affect a 
variety of age groups. They are also conditions in 
which the role of medical care is fairly clear. Clinical 
and epidemiologic data for these condtions and 
procedures were also collected. 

PSROs were selected for this study according to 
the length of time they had been conditional, the 
number of audits and reaudits they had reported 
through the PMIS, and the overall quality of their 
MCE activity as judged by the investigators and 
HSQB. Of the fourteen eligible PSROs, eight agreed 
to participate in the study. 

The health benefits associated with the observed 
variation rates and VR changes were estimated by 
consensus among a panel of physician experts. For 
each of the diagnoses and procedures, the panel 
was asked to predict patient outcomes over the 
five years following hospitalization, for four levels of 
care: no care, care as indicated by the audit, care 
as indicated by the reaudit, and ideal care. The 
predictions were made by estimating the number of 
months over the five-year period that a hypothetical 
patient would spend at each of six levels of health 
status, ranging from complete wellness to death. 
For comparison, the panel was provided with data 
on the average time spent at each of the six levels 
of health status during a five-year period for the U.S. 
population of the same age as patients hospitalized 
for each of the four diagnoses and procedures. Panel 
estimates were discussed until a consensus was 
reached, and verified by consultation with twelve 
additional experts. 

The resulting judgements were further combined 
into a weighted index of "health status months" 
(HSMs). The number of months spent at each of the 
six health status levels was weighted with a value 
ranging from (death) to 1 (perfect well-being), 
using values established by Kaplan, Bush and Berry.^ 
The result of the MCE benefit analysis, then, is an 
index expressing both quality and quantity of life in 
terms of health status months. 

The study then estimated the monentary value of 
the following components: 

• dollar value of health benefits, based upon 
studies by health economists which valued a 

a "healthy month" at between $500 and $1000. 

• costs of MCEs, based upon physician and non- 
physician hours utilized, hourly salary and wage 
rates, and non-labor and overhead rates. 

• costs of care changes from audit to reaudit, 
based upon actual charges for ancillary procedures 
and changes in utilization of these procedures 
from audit to reaudit 



'•'R. Kaplan, J. W. Bush, and C. C. Berry, "Health Status: 
Types of Validity and the Index of Well-being," Health 
Services Research (Winter 1976), pp. 478-507. 



5.6.3 Findings 

Figure 24 shows the results of the physician 
panel's estimation of HSM values. For each diagnosis 
or procedure, the first bar indicates the expected 
health status months for the U. S. population of the 
same age as hospital patients. The other bars 
represent, in turn, health status months for no care, 
initial audit care, reaudit care, and ideal care. 
Differences between the bars can be interpreted as 
the relative health benefits (in HSMs) of each mode 
of care, e.g., in congestive heart failure, care as 
indicated by the initial audit would yield 9.02 HSM 
(i.e., 25.63-16.61) more than no care, while care 
indicated by the reaudit would increase HSMs by 1.39, 
and ideal care would further increase HSMs by 2.97. 
In this case, the MCE impact (the improvement of 
care from audit to reaudit) represented 32 percent of 
the health benefit that would be obtained by moving 
from audit care to ideal care. It is apparent that 
congestive heart failure shows the most MCE impact 
of the four topics, partly because the gap between 
initial audit care and ideal care for the others is so 
small- 
Table 73 presents the dollar values of costs and 
benefits, and the benefit-cost ratios obtained for the 
four MCE topic areas. Columns (1) and (2) present 
estimates for the five year period following 
hosDitalization of the annualized monetary value of 
MCEs in the four topics, calculated by multiplying 
MCE impact by the number of patients at risk in the 
audit sites in the course of a year and the dollar 
value of one HSM. The high and low estimates are 
derived by valuing an HSM at $1000 and $500, 
respectively. Columns (3) and (4) present high and 
low estimates of the costs of performing the sampled 
MCEs, based on physician and non-physician hours 
and wages and two assumptions about the ratio of 
nonlabor and overhead costs to labor costs. Column 
(5) shows the increased treatment costs attributable 
to MCEs, based on the increased numbers of ancillary 
procedures used at reaudit and their unit costs. 
The data are summarized in two ways: as net 
benefits by taking the difference between benefits 
and costs under both high and low assumptions, 
and as benefit-cost ratios. The latter are calculated 
two ways: (1) by comparing high benefits to high costs; 
and (2) under a conservative "minimax" assumption 
which divides the low benefit estimate by the high 
cost estimate. By either method, the benefit-cost 
ratios are substantially greater than one, i.e. benefits 
exceed costs by a wide margin. 

5.6.4 Discussion 

It is evident that the data and the findings of this 
study are separated by several layers of assumptions 
and approximations. In addition, no attempt was 
made to sample MCEs and PSRO hospitals repre- 



124 



1 

10 

Congestive Heart Failure 



Figure 24 

Health Status Months In Five Care Categories: 
Mean Estimates of Physician Panel 

Health Status Months (Average Patient) 



20 



T 

30 



Bar 1 |U. S. Average, Age 74 



39.94 



Bar 2 |No care 16.61 I 



Bar 3 


[initial Audit 


25.63 1 








Bar 4 


iReaudit 


27.02 1 








Bar 5 


1 Ideal Care 


29.99 1 



Asthma 



T 

40 



50 



I 
60 



MCE impact=1.39 HSM 



Bar 1 


|U. S. Average, Age 42 


52.69 1 








Bar 2 


|No care 


45.63 1 








Bar 3 


1 Initial Audit 


50.17 1 






^ 


Bar 4 


iReaudit 


50.58 1 








Bar 5 


1 Ideal Care 


52.45 1 


Cholecystectomy 


Bar 1 


|U. S. Average, Age 52 


50.48 1 








Bar 2 


|No care 


48.07 1 








Bar 3 


1 Initial Audit 


49.90 1 






1^ 


Bar 4 


IReaudit 


50.13 1 








Bar 5 


1 Ideal Care 


51.10 1 


Cesarean Section (mother only) 


Bar 1 


|U. S. Average, Age 25 


53.48 1 








Bar 2 


|No care 


46.22 1 








Bar 3 


1 Initial Audit 


54.03 1 








Bar 4 


IReaudit 


54.26 1 








Bar 5 


1 Ideal Care 


54.82 1 



MCE impact=.41 HSM 



MCE impact=.23 HSM 



MCE impact=.23 HSM 



125 



(0 

U 



0) 

m c 

< " 

o 
O 



(0 

E 

E 

a 

CO 



X I lO 

.£ 0) o cc ^ 
2DQ i^ 



X Q) O rr -«. 

CO :r 



0) lO 

o c ^v 
CO JL 

c 
m 

X '-^ T 

LU -i- 



«= c £ 2 
O "^^ 



(1) 



CO 
o 
O 

UJ 

o 



t^ CO 00 CO o> 

to CO iri CO td 



iq lo h- in 00 

CO CD T^ CO CO 
CO •.- T- 



<0 CO CVJ T- CM 

00 00 en o> lo 
w CD r-_ CD in 

to in in cm" of 

en Tt O CM CD 
CD CO CM CM 



</> 



</> 



CO 



O m 

QX 



in 



5 "_ 

O C TT 

_j -^ >— 

V) 

LU 



x: CO 
9>Eco 
X ■*= ^-' 
-^ CO 
LU 



C3) 1- T- CD I^ 

C3) T- 1- in t^ 
o in ^f CD CO 

cm" T-" OO" CJ>" T-" 

o) O) T— ^ in 
CO CD ^f in 



«» 



00 
CD 



c/> 



CM 



1^ C\J o 

in CO o 

-r-OOi 

in co"c3f 



0) 
. CO ^^ 

! E CM 

■*: ^-' 
CO 
LU 



CO O O T- TJ- 

CO CD CD 1- C3) 
N-_^ Cn O^ C0_ 0_ 
0>" OO" h«-" o" CO" 
1- CM Tl- C3> 



o in o o tn 
00 t^ m 1- 7- 

CD in ^ •^ ID 
T-" CO" cm" CJ)" cd" 

'a- T- in N. 00 



e» 



tfl^ 



o CO C3> ^r CO 
CO Tf o CO ^r 

•^O CD C3 m 

h-" ^r co" CD in 
T- m CO CO N- 

^- CO CM CO 



0) 

s: "co 
.2>E: 

^ CO 
LU 



<^ 



</> 



O CD 00 00 CM 
CO 00 1- CM CD 
OC)^ O <D T-_^ O 
^ 00 CD cm" T-~ 

CO o h- CO in 

■* T- CD m N- 



<«• 



CM 



o 
'q. 
o 







>. 








E 








o 
















o 








<0 








tE 








2 




(0 


O •■£= 


o 




E 


<» s? 


t- 


Li. 

X 


"tn 


1^ 




o<oo 





126 



sentatively. Accordingly, the study should be regarded use of these numbers suggests that for some MCEs 

as a first attempt to answer some difficult questions benefits far exceed costs. Given these results, further 

about the effectiveness of MCEs in relation to their research w^hich samples more representatively, 

cost. The study is clearly exploratory rather than supplies more exact data, and replaces some of the 

definitive. The method employed is one often used in subjective judgements and assumptions with facts, 

situations of uncertainty: the combination of data is likely to conclude that quality assurance in general, 

with expert judgement and explicit assumptions. and MCEs in particular, justify the expenditures 

While the benefit-cost estimates are admittedly invested- 
soft, it is encouraging to see that even conservative 



127 



6. Profile Analysis 

PSROs have a unique resource in that they are 
mandated to collect information on each hospital 
stay involving Federal patients, and to analyze the data 
to detect problems in the utilization and quality of 
care in their areas. The latter activity is commonly 
known as profile analysis. This chapter presents a 
definition of profile analysis, technical considerations 
affecting its use, and a discussion of the potential role 
of profiles in the PSRO program. The actual status 
of profile analysis in the program is reviewed by 
examining (a) the activities undertaken by HSQB in 
support of profile analysis, (b) a set of reports from 
the PSROs describing profiles conducted in the first 
half of 1979, and (c) the findings of a series of case 
studies which examined the factors which facilitate 
the effective use of profiles in seven PSROs. 



6. 1 The Mandate for Profile Analysis 

Of the three principal components of the PSRO 
hospital review system, only profile analysis is 
specifically set forth in the law. Under Section 1155 
of P.L. 92-603, entitled "Duties and Functions of 
Professional Standards Review Organization," para- 
graph (a)(4) reads: 

(4) Each Professional Standards Review Organiza- 
tion shall be responsible for the arranging for 
the maintenance of and the regular review of 
profiles of care and services received and provided 
with respect to patients, utilizing to the greatest 
extent practicable in such patient profiles, methods 
of coding which will provide maximum confiden- 
tiality as to patient identity and assure objective 
evaluation consistent with the purposes of this 
part. Profiles shall also be regularly reviewed on an 
ongoing basis with respect to each health care 
practitioner and provider to determine whether the 
care and services ordered or rendered are consistent 
with the criteria specified in clauses (A), (B), and 
(C) of paragraph (1). 

The reference above to paragraph (1) pertains to the 
basic goals of PSROs: i.e.. to determine that services 
provided (a) are medically necessary, (b) meet 
professional standards of care, and (c) are provided 
at the appropriate level of care. Beyond this, the 
law is not specific; the translation of the mandate 
to a system of review requires a multitude of decisions 
and actions at both the national and PSRO levels. 



6.2 The Nature of Profile Analysis 

Definition. The term "profile", in the PSRO 
context, refers to the presentation of aggregated 
data in formats which display patterns of health care 
services over a defined period of time. Although not 



every display of data can properly be classified as a 
profile, there is considerable latitude for local 
innovation in developing profiles- What is common 
to all profiles, however, is the use of automated data 
systems based on patient record abstracts by the 
PSRO in support of PSRO objectives. This differentiates 
PSRO profiles fundamentally from the private hospital 
discharge abstract reports which preceded them. 
Uses. Profile analysis provides the PSRO with 
information on patterns of hospital utilization and 
quality of care for Federal patients. Goran et al.^ 
listed the following uses in 1975, before many PSROs 
had achieved profiling capability: 

1. To help manage review activities within a 
hospital (regardless if hospital or PSRO 
performs review). 

• Concurrent Review (OR) — assist in determing 
priorities and areas of exemption for OR 
activities; monitor effectiveness of previous 
OR efforts. 

• Medical Care Evaluation (MCE) Studies — 
assist in determining priorities for MCE 
studies. 

2. To help manage review activities within a PSRO. 

• Understand patient volumes and characteristics 
of hospitals within area. 

• Aid in overall PSRO evaluation. 

• Aid in monitoring impact of hospital or 
PSRO review effects- 

• Enable comparisons of utilization trends in 
similar hospitals (e.g., shifts in number of 
admissions, days of stay, or rates of diagnoses 
and procedures). 

• Aid in assessing impact of specific PSRO 
activities. 

• Provide information for comparisons of 
PSRO area statistics with other PSRO areas, 
for development of LOS norms and standards 
for CR activity, or standards for subsequent 
profile analysis. 

3. To assist DHEW with program management 
and evaluation. 

Types of Profiles. The PSRO legislation specifically 
identifies three types of profiles which must be 
reviewed: (1) patient profiles, (2) practitioner profiles, 
and (3) institutional profiles. This represents the 
minimum set of profiles which must be prepared. 

1. Patient profiles 

The law refers to profiling the "care and 
services received and provided with respect to 
patients," which, for the hospital setting, is 
generally interpreted as conducting analysis on 
groups of patients. These analyses group patients 
according to diagnosis, procedure or therapy, 
payment source, or by demographic charac- 



' M. Goran, J. Roberts, M. Kellog, J. Fielding, and W. Jessee, 
'The PSRO Hospital Review System," Medical Care 13, 
no. 4 (Supplement, April 1975). 



128 



teristics, such as age, sex, race, or residence. 
These types of groupings are the basis of 
practitioner or institutional profiles, and are also 
generated across institutions for the entire 
PSRO. Profiles are also produced by HSQB in 
order to assist the PSROs in identifying problems. 
Profiles can be generated on the readmission 
patterns of individual patients, looking at 
admissions both to the same hospital and to 
other hospitals in the PSRO area. The require- 
ment in the PSRO data set for a unique identifier 
facilitates these types of analyses, but, as 
with all PSRO data analysis, careful consideration 
must be given to maintaining strict con- 
fidentiality as to patient identity- 

2. Practitioner profiles 

Profiles are to be prepared for each health 
care practitioner who is providing inpatient 
health care services to Federal beneficiaries. 
Since the hospital discharge data collected 
by the PSROs include a unique identifier for 
each attending and operating physician, it is 
possible to profile a physician's practice both 
within a hospital and across hospitals in the 
same PSRO area. 

Practitioners can be profiled on lengths of stay, 
admission patterns, incidence of surgery, and 
concurrent review patterns, such as denials of 
continued stay certifications. To the extent 
that the data are available from optional items 
collected by the PSRO or from MCE studies. 
It may also be feasible to profile practitioners 
using the criteria established for the process 
of care delivered to patients. Although some 
overall statistics may be useful for volume 
assessment, the usefulness of practitioner 
profiles is enhanced if case-mix and severity 
of illness are taken into account. 

3. Institutional profiles 

institutional profiles identify problem areas, 
monitor review activity, and assess performance 
in both delegated and nondelegated hospitals. 
These profiles are produced for each hospital 
under review in the PSRO area. In addition to 
profiles on individual hospitals, larger PSROs 
may choose to produce profiles on groups of 
hospitals clustered by such factors as teaching 
status, size, presence of specialized services or 
facilities, delegation status, or volume of 
Federal discharges. 

Again, It will be essential in certain of these 
profiles, such as those examining lengths of stay 
or utilization patterns, to control, wherever 
possible, for such patient-mix characteristics 
as diagnosis, age, and presence or absence 
of surgery. 

Data. The primary data base used by both PSRO 
and the central office consists of discharge abstracts 
prepared in the hospital for each Medicare, Medicaid, 



and Maternal and Child Health patient. The abstracts 
contain elements specified in the Uniform Hospital 
Discharge Data Set, optional elements determined 
by each PSRO, and, until recently, additional 
elements which describe the review process. Abstracts 
are delivered by the hospitals to a data processor 
which has contracted with the PSRO to prepare 
data tapes and routine reports. Copies of these tapes 
are delivered quarterly by the PSROs to the Division 
of Data Planning and Analysis of HSQB- The system 
is generally known as the PSRO Hospital Discharge 
Data Set (PHDDS). 

Other sources of data may be used when PHDDs 
data are lacking or where local opportunities present 
themselves. MEDPAR data derived from Medicare 
claims files must be used, for example, if a PSRO 
wishes to compare pre-PSRO experience with current 
data, since PHDDS data do not exist for the pre-PSRO 
period nor for a considerable time after the 
establishment of most PSROs. Commercial abstract 
services may provide comparable information. Some 
PSROs have also used Medicaid Management 
Information Systems (MMIS), where they exist, to 
study utilization patterns in ambulatory or long term 
care. These sources may also be the primary ones 
for a PSRO which wishes to analyze profiles before 
its own data system is operational. 

Analysis of Profile Data. Profile analysis is a 
monitoring and problem finding activity. The methods 
used to analyze the data are as diverse as human 
ingenuity can make them, but certain common pro- 
cedures can be observed. Generally, the kinds of 
comparisons that a PSRO will make are: 

• Comparison of patterns of care by similar 
providers; 

• Comparison of current patterns with previous 
patterns; 

• Identification of patterns that deviate from 
established norms, criteria and standards; or 

• Tracking of care provided to particular patients 
or by particular practitioners within the entire 
PSRO area. 

The clearest single statement of the process of 
profile analysis is contained in Goran's recent article 
in Medical Care. The analysis of utilization data 
is described in four sequential steps: 
"1) Profile analysis is used to define aberrant 
practice patterns and to help select from those 
aberrant patterns the ones most likely to 
represent problems in the delivery of health 
care services that are correctable through PSRO 
review; 

2) Profile analysis, in conjunction with other 
available techniques, is used to help search 
for the cause of the aberrant patterns that 
represent potentially correctable problems; 

3) Profile analysis leads to the selection of 
priorities for the PSRO objective-setting process 
and helps to make reveiw more problem- 
oriented; and 



129 



4) Profile analysis, in conjunction with other 
techniques, is used to monitor the effect of 
corrective actions taken by the PSRO and other 
bodies to help solve the priority problems." ^ 

The first step, the detection of aberrant patterns, 
is the most crucial. Beginning at a high level of 
aggregation, analyses become more specific as 
problems are detected. A PSRO may first compare 
its own distribution on a given parameter (e.g., average 
length of stay for acute myocardial infarction) with 
national and regional distributions of PHDDS data 
prepared by the HSQB. The objective at this stage of 
the analysis is to detect a local practice pattern that 
deviates from national, regional, or PSRO norms. 

Within a PSRO, comparisons are often made between 
all hospitals, but for specific purposes only hospitals 
of a certain type (e.g., size) may be compared. 
These comparisons may extend to hospitals in other 
PSROs to provide sufficient numbers of cases. 
Whfen a problem is detected at the hospital level, 
physician profiles may then be compared to see 
whether the problem is specific to certain practitioners. 
Physician profiles are also produced at the PSRO 
level, especially when practitioners admit patients 
to a number of area hospitals. For some specialties, 
it may be necessary to extend comparisons beyond 
the PSRO to obtain sufficient case frequencies. 

The parameters most frequently examined include 
days of care rates, admission rates, procedures 
rates, average length of stay, pre and post-operative 
length of stay, and proportion of long and short 
stays. The length of stay variables are frequently used 
because they relate days of care in the numerator 
to number of admissions in the denominator, and not 
to populations outside the hospital, which are often 
difficult to measure. Pre or postoperative ALOS 
similarly relates pre/postoperative days of care to 
the number of surgical admissions. These are thus 
institution-based rates and do not rely on the size of 
the population at risk for hospitalization. 

Other parameters of utilization, such as days of 
care rates, admission rates, or procedures rates, are 
calculated as rates per thousand in the eligible 
population. These are often more informative than 
average length of stay, since they relate more closely 
to health care expenditures, but they do require 
that the size of the eligible population be known. 
For this reason, rate analysis can be performed only 
at the level of the PSRO or larger areas, except in 
special cases like HMOs or geographically isolated 
hospitals where the service population denominator 
is known. The Medicare enrolled population can 
easily be obtained for each PSRO, even when adjust- 
ments for migration amonq PSROs must be made. 
However, the number of Medicaid eligibles can be 
reliably estimated for only a few states. 



'M. Goran, "The Evolution of the PSRO Hospital Review 
System," Medical Care 17, no. 5 (Supplement, May 1979), 
p. 18. 



Standards used in analysis of profile data are 
frequently established empirically. It is usually not 
certain, given the current status of medical knowledge, 
which patterns represent the best care. Instead of 
having absolute standards derived from the scientific 
literature or from authoritative consensus, the 
profile analyst must often assume that that pattern of 
care which is most prevalent is best. Some PSROs 
have attempted to build measures of quality of care 
or severity of illness into their profile analyses in 
order to strengthen the validity of this assumption. 

Deviation from the empirical norm is thus used as 
presumptive evidence of deviation from desirable 
patterns of care. The profiles done at the national 
level by HSQB, for example, identify those PSROs 
which are 20 percent above or below the national 
average on selected parameters of care. PSROs 20 
percent or more above the norm are asked to use 
the objective-setting process to bring their average 
down. Those 20 percent or more below the norm are 
asked to examine their profiles carefully to assure 
that their low average does not conceal problems 
in care, and then to consider relaxing concurrent 
review in these areas. 

The process of applying standards of care in 
analyzing profile data is thus far from automatic. Abso- 
lute standards, in most cases, do not exist. The 
process is one of successive approximations, com- 
bining skills in data manipulation with clinical 
interpretation. As PSROs gain experience in profiling, 
they are able both to revise the standards they use and 
to identify areas in wihch scientific validation is 
needed. 

Potential role of profile analysis in the PSRO 
hospital review system. Profile analysis is usually 
the last of the three components of the PSRO review 
system to be implemented in each PSRO, largely 
because the technology and skills needed take time 
to acquire, but also because it is regarded as more 
important first to establish the basic mechanisms of 
concurrent review and medical care evaluation. 
Profiling is receiving increasing emphasis now that 
the basic mechanisms are established in most areas 
and the program is shifting to problem-oriented review. 
The philosophy of the program is that, once imple- 
mented, profile analysis can fundamentally alter the 
review system, making PSROs more responsive to 
problems in health care delivery at the local level, 
and enabling them to respond to resource constraints 
by reducing review costs. In particular, at a time 
when static appropriations, rising costs, and pressure 
to demonstrate effectiveness are encouraging PSROs 
to replace labor-intensive 100 percent utilization 
review with problem-focused review, it is likely that 
data analysis will increasingly be relied upon to direct 
the allocation of effort. For most PSROs, however, 
the centrality of profile analysis is potential rather 
than actual at this time- 



130 



6.3 Program Development 

Profile analysis had some antecedents in the 
activities of private hospital discharge abstracting 
services and some medical care foundations, but the 
scope and scale of data analysis in the PSRO program 
is essentially unprecedented. This section describes 
the reasons for the slow diffusion of profiling 
techniques among the PSRO and the steps taken by 
HSQB to support and foster it. 

Early development. While utilization and quality 
reviews rely upon a fairly simple technology, profiling 
requires a complex system of data acquisition and 
processing and sophisticated skills of analysis. 
During the first two and one-half years of the PSRO 
program, primary emphasis was placed on the 
recruitment of membership, development of the 
PSROs' organizational structure, and the development 
and implementation of concurrent utilization review 
plans. Once concurrent review was established, 
attention turned to implementing quality of care 
reviews, through medical care evaluation studies. At 
this time, PSROs were also setting up data systems 
that would support review activities and allow them 
to meet their reporting requirements. Few PSROs 
were able quickly to establish systems to collect and 
process data of sufficient quality to perform profile 
analysis. Neither the technical nor clinical skills 
needed were readily available. Among the exceptions 
were those PSROs which had developed data 
capabilities as Exoerimental Medical Care Review 
Organizations (EMCROs). As a result, the profile 
analysis component of the PSRO review system was 
the last to develop. Some other factors are related 
to this slow development: 

• Profile analysis was a relatively new type of 
medical care review. Unlike concurrent review 
and MCEs, there was little or no expertise in this 
area of review from which to draw. Therefore, 
PSROs interested in developing an effective profile 
analysis system had few sources of technical 
assistance, training, or successful models. 

• During the early years of the PSRO program the 
central office did not emphasize profile analysis. 

• The paucity of existing data both on the local 
and national levels deterred efforts to develop 
meaningful approaches to review based on 
profiles. 

• Start-up problems are inherent in any new data 
system. These relate both to collecting data and 
having it processed in a useful manner. 
Consequently, PSROs were not able to provide 
HSQB with data of a quality or quantity which 
would enable it to produce national PSRO 
HosDital Discharge Data Set (PHDDS) profiles 
until 1978. PSROs and HSQB staff encountered 
difficulties in obtaining national data in a format 
that was suitable to producing PSRO area profiles. 

• During the early years of the PSRO program, 
there were few qualified data analysts on the 



staffs of most PSROs, nor were physicians 
sufficiently trained to use the data or to have 
confidence in it. 

Program initiatives. Pressure since 1977 to increase 
the cost-effectiveness of review has encouraged the 
program to devote additional attention to profile 
analysis. The overall strategy has been to develop 
technical assistance manuals which are available to 
all PSROs, while those PSROs which have shown the 
most initiative in using data are offered special grants 
to progress further. Thus, while there has been a 
good deal of central guidance of the growth of 
profile activity, the diffusion of the art has depended 
equally upon communication among the PSROs. 

Key events in the chronology of initiatives undertaken 
by HSQB include: 

• Designation of eight PSROs as AUTOGRP sites 
(1977). AUTOGRP ' is an interactive computer 
system developed at Yale University specifically 
for the analysis of health care data. 

• Special grants of $30,000 to seven PSROs 
which were advanced in profile analysis, to be 
used to enhance their profile activity as they saw 
fit (September, 1977). 

• On-site technical assistance to 14 PRSOs (1978), 
later extended to a total of 34 PSROs (1979). 

• Publication of semi-annual national profiles for 
all PSROs which had submitted PHDDS data tapes 
(beginning July, 1978). These permitted com- 
parisons at the regional and PSRO level. Areas 

in which each PSRO's performance was 20 percent 
above or below the national norm were noted. 
The national profiles served both to demonstrate 
to PSROs fundamental techniques of profile 
analysis and to show that the patterns were 
observed at both regional and national levels. 

• Reoional workshops for all PSROs and a 
national conference of advanced PSROs( 1979). 



6.4 Description of Profile Analysis Activity 

Until recently, little systematic information was 
available regarding the profile analyses actually being 
done in the PSROs. New reporting requirements were 
instituted in January 1979* which provide the central 
office with quarterly reports from the PSROs on each 
profile analysis conducted. This section uses the 
first two quarters' reports to describe the number 
and types of profiles performed in the first half of 1979. 

6.4.1 Data 

Profile Analysis Activity Report — Form HCFA 141. 
The data source for this review is a routine form 



^ R. Mills, ef a/., "AUTOGRP: An Interactive Computer System 
for the Analysis of Health Care Data," Medical Care 
(July 1976). 

' "Revised Federal Reports Manual," PSRO Transmittal 
No. 87, January 2, 1979. 



131 



which provides the nine data elements shown in 
Table 74 for each profile: 

TABLE 74 
Form HCFA 141 Data Elements 



PSRO number 
PSRO name 
Quarter ending date 
for the report 
submitted 
Title of profile 
Profile analysis 
number 



Phase of profile 
analysis 
Data source 
Person-hours utilized 
by PSRO staff, physi- 
cians and consult- 
ants 

Action taken as a 
result of profile 



Space is provided on the report for narrative 
discussion to expand on any of the items and/or to 
add other pertinent information about the activity. 
The narrative section enables the PSRO to expand 
upon such areas as the purpose of the profile, an 
explanation of the findings, any action taken, the 
method for monitoring the results of the action taken, 
and a description of any problems encountered. 
The use of this narrative section by PSROs varies 
greatly from no comments to extensive discussion 
including findings, display formats, etc. The HCFA 141 
currently represents the only source of written 
information for assessing the status of profile analysis 
activity on a national basis both in terms of quantity 
(how much profile analysis is being done) and quality 
(what types of profile analysis are being conducted 
and how well). 

Limitations to tlie Interpretation of HCFA 141 Data. 
In providing an overview of profile analysis across 
PSROs based on an analysis of the HCFA 141, it is 
important to recognize some of the limitations of 
this assessment. The fact that profile analysis is a 
newly implemented activity and the HCFA 141 is a new 
form results in a wide variation in the way profile 
analsyis activity is reported. Specifically, in interpreting 
the data obtained from the HCFA 141, the following 
should be taken into consideration: 

1. This evaluation of profile analysis activity is 
based only on HCFA 141s submitted for the 
first and second quarters of 1979. Consequently, 
the data may be insufficient for characterizing 
the nature of profile analysis activity, and the 
amount of activity reported may not accurately 
reflect the actual amount of profile analysis 
being conducted by PSROs. 

2. While nine specific items are requested on 
the HCFA 141, there is a great disparity across 
PSROs in terms of the amount of information 
supplied and the manner in which it is reported. 
Even the definition of the unit of analysis is 
inconsistent: a study carried out in several 
stages may be reported as one or several 
profiles. The dependence of the evaluation on 



the HCFA 141, and the apparently inconsistent 
manner in which the reporting form is used, limits 
the value of analysis. 
3. Since much of the information on the HCFA 141 
is in narrative form, it is difficult to abstract it 
in an entirely comparable manner. While a 
standard protocol was used for abstracting the 
information from the HCFA 141s, some arbitrary 
decisions were necessary to categorize some of 
the information. Every effort was made to make 
the analysis as consistent as possible. 

Despite these limitations, a descriptive assessment 
of the status of profile analysis activity can be made, 
and should be useful in developing a better under- 
standing of the amount and type of profile analysis 
presently being conducted by the PSROs. 

6.4.2 Findings 

Using a protocol to abstract information from the 
HCFA 141s, evaluators identified the following 
significant characteristics: 

• Number of PSROs reporting profile analysis 

• Data sources used to conduct profile analysis 

• Types of profiles being analyzed 

• Reasons why profile was conducted 

• Types of measures used to report profile findings 

• Actions taken as a result of profile findings 

• Types of impact reported by PSROs 

Number of Reporting PSROs. Of the 188 PSROs 
active during the first half of 1979, 123 reported 
profile activity. An additional 22 PSROs completed 
the reporting form but indicated that they had not done 
profiling during the period, while 43 PSROs did not 
submit the form. The 123 PSROs which reported 
profile activity reported a total of 911 profiles or an 
average of 7.4 profiles per reporting PSRO. 

Aae of PSRO. PSROs reporting profile activity 
tended to be older. Table 75 shows the date of 
conditional status for all active PSROs, and for those 
which reported profiles. Seventy-six percent of 
PSROs which became conditional before 1978 
reoorted profiles, as opposed to 54 percent of those 
which attained conditional status in 1978 and none 
of those which began activity in 1979. 

TABLE 75 
PSROs Active in Profiling, by Age 







Number 


Percent 


Time 


Number 


of 


of 


PSRO 


of 


PSROs 


PSROs 


Became 


Active 


Reporting 


Reporting 


Conditional 


PSROs 


Profile 


Profile 






Activity 


Activity 


Before 1978 


123 


93 


75.6% 


1978 


56 


30 


53.6% 


1979 


9 








All 


188 


123 


65.4% 



132 



Size of PSRO. Larger PSROs are more likely to 
perform profile analysis. Table 76 shows the number 
of Federal discharges for ail PSROs active during 
the period and for those reporting profiles. 

TABLE 76 
PSROs Active in Profiling, by Size 







Number 


Percent 


Estimated 
Number of 


Number 
of 


of 
PSROs 


of 
PSROs 


Federal 


Active 


Report- 


Report- 


Discharges 


PSROs 


ing 
Profile 


ing 
Profile 






Activity 


Activity 


> 100,000 


31 


26 


83.9% 


50,000 to 100,000 


38 


32 


84.2 


20,000 to 50,000 


85 


49 


57.6 


< 20,000 


34 


16 


47.1 


All 


188 


123 


65.4% 



Data Used in Profiles. PSROs are requested to 
identify on their HCFA 141s the data sources used 
in conducting profile analysis. Of the 123 PSROs that 
reported profile analysis, almost 75 percent reported 
using PHDDS data. 

Existing data systems, which include MEDPAR 
and MM IS information, were cited as the next most 
frequently used. Table 77 identifies the percentage 
of the 123 PSROs that reported profile analysis activity 
using each data source. Since PSROs may use more 
than one data source in analyizing one profile, the 
percentages exceed 100. 

TABLE 77 
Data Sources Used for Profiles 



Data Source 



Percent of PSROs 
Reporting Use 



PHDDS 

Existing data systems 
Other components of the PMIS 
Baseline data on pre-PSRO 
experience 



71.5% 
57.9% 
21.9% 

24.7% 



Types of Profiles. In abstracting information on 
the types of profiles being analyzed, profiles were 
assigned to the following main categories: 

Descriptive — profiles that are either a listing, 
grouping or identification of individual variables 
(e.g., "Number of readmissions at Hospital X") 

Analytic — profiles that undertake a comparative 
analysis between two or more variables (e.g., 
"Analysis of LOS, by diagnosis, at Hospital X") 



Review Process — profiles that analyze or evaluate 
some aspect of the review process, such as number 
of denials, extensions, or grace days 

Data Quality — profiles designed to monitor data 
quality 

Other — profiles that do not fall under the above 
categories. Some examples: 

"Analysis of non-covered days" 

"Analysis of skilled and intermediate care days 
by hospital" 

"Calculation of Utilization Index" 

"Same Day Stay Study" 

Unknown — those profiles which are reported in a 
manner which is unclear, vague and precludes 
assignment to categories 

Table 78 shows the percentage of the 123 reporting 
PSROs which analyzed each type of profile, and the 
percentage of the 911 profiles which were of each 
type. Since most PSROs analyzed more than one 
type of profile, the percentages by PSRO total more 
than 100. 



TABL 


.E 78 




Types of Profiles 






Percent 

of 

Reporting 

PSROs 


Percent 


Type 


of 
Profiles 


Descriptive 






generate baseline data 


4.1 % 


0.7% 


diagnostic/procedure 


30.1 % 


12.7% 


hospital 


46.3% 


22.8% 


physician 


22.8% 


6.0% 


other 


21.9% 


7.2% 


Analytic 






diagnosis by hospital 


28.5% 


16.9% 


diagnosis by physician 


0.8% 


0.1 % 


diagnosis by hospital 






by physician 


11.4% 


3.0% 


procedure by hospital 


13.8% 


6.0% 


procedure by physician 


3.3% 


0.5% 


procedure by hospital 






by physician 


5.7% 


2.0% 


Review Process 


26.8% 


10.0% 


Data Quality 


4.1% 


0.7% 


Other 


29.3% 


11.0% 


Unknown 


1.1% 


0.3% 



100.0% 

The fact that a large percentage of the profiles 
reported were at the hospital level of analysis is 
consistent with what might be described as a natural 
progression in the development of a PSRO's profile 
analysis capabilities. As PSROs undertake profile 
analysis, their first efforts are centered on using 



133 



descriptive profiles to better understand utilization 
patterns and/or identify potential problem areas. 
Having completed this initial phase of activity, PSROs 
begin to use profiling techniques that are more analytic 
in nature, and which focus primarily on the review 
process or analysis of utilization by diagnosis or 
procedure. Finally, PSROs that have gained both 
confidence and sophistication in their profiling 
techniques conduct analytic profiles that are con- 
centrated at the hospital level of analysis and 
subsequently at the physician level. 

Reason for profiles. In analyzing the purpose of 
profile activities reported by the PSROs, a methodology 
was developed to categorize the stated "reasons 
for profile". In many cases, a PSRO indicated more 
than one reason for performing its analysis. Of the 
123 PSROs that reported profile analysis for the study 
period, over 50 percent stated that one or more 
profiles were analyzed to identify or evaluate variables 
(e.g., LOS, admission rates) that represent variations 
from norms or standards (e.g., PAS,^ state/regional 
norms, PHDDS reports). The other major reasons for 
undertaking profiles cited by PSROs were to monitor 
the review process, to identify potential problem 
areas, and to focus review (in or out). The PSROs 
also cited "other" reasons for their profiling activities 
(see examples below). 

Of the total number of profiles reported in each 
quarter. Table 79 identifies the percentage of 
profiles in each "reason" category: 



TABLE 79 
Stated Reasons for Profiles 



Reason 



Percent of 
Profiles 



Monitor review process 
Provide baseline data/reference 
Identify topic for MCE 
Identify potential variation 

from norm 
Objective setting /problem 

identification 
Focusing review 
Other 



Some examples of "other" reasons cited by 
PSROs include: 

"Determining cost of administrative days" 
"Develop population-based admission rates" 
"Identify nature of short stays" 



10.6% 


4.1% 


1.6% 


29.3% 


17.9% 


17.9% 


18.0% 



^ PAS=Professional Activity Study, a hospital data abstracting 
program of the Commission on Professional and Hospital 
Activities (CPHA), Ann Arbor, Michigan. 



"Determine number of days patients await 

placement in LTCF" 
"Identify health care planning needs." 

As indicated by the table, the main reasons cited 
for performing profile analysis were to identify 
potential variations from norms, to set objectives and 
identify problems, and to focus review. To a large 
extent, these reasons overlap and interrelate with 
each other and it seems likely that many PSROs 
conduct profile analysis to achieve these objectives 
simultaneously. Since it is difficult to assess the degree 
to which reasons overlap, the percentages presented 
may be somewhat misleading. 

Findings of Profiles. An exhaustive list of types of 
findings are reported as a result of PSRO profiling 
activities, so that it was necessary to develop a 
methodology that would permit aggregation of the 
data. It was difficult, however, for evaluators 
abstracting this information to develop a meaningful 
interpretation. A large percentage of PSROs did not 
indicate what the findings of their studies were. It 
was impossible to tell whether the fact that no 
f'ndings were reported on the HGFA 141 indicated 
that (a) the PSRO was still in the analysis or corrective 
action phase of its profiling, (b) there were no 
findings, or (c) the failure to document findings could 
be attributed to poor reporting. Of the PSROs that 
did report findings, a majority indicated a variation 
from established norms and standards relating to 
utilization and/or quality of care. Some of the more 
frequently cited dimensions on which such variations 
were found included: length of stay, pre-operative 
length of stay, admission rate, readmission rate, 
mortality, nursing home transfer rate, short stays, and 
weekend admissions. 

Actions fallen. An important aspect of the 
evaluation was to identify the types of actions 
reported by PSROs as a result of their profile analysis 
activities. Many of the actions taken were in the 
form of a request for additional information either from 
hospital staff or UR committee and/or continued 
analysis by PSRO staff. This may indicate that the 
analysis was still in progress and that more definitive 
actions mav be reported subsequently. As with the 
reporting of findings, a large percentage of PSROs 
did not report any actions taken for some of the 
profiles analyzed. It is not clear from the reports 
whether additional analysis was undertaken, or that 
despite profile findings the PSRO failed to take any 
concrete actions, or as before, that the problem was 
in poor reporting. 

Of the 123 PSROs that reported profiled analysis 
activity, almost one-half stated that their profile 
findings had resulted in either a "focusing in" 
(intensification of utilization review) or a "focusing out" 
(automatic certification) for certain physicians, 
hospitals, procedures, diagnoses, etc. Table 80 
identifies the percentage of PSROs as well as the 
percentage of profiles which reported each type of 
action. 



134 



TABLE 80 
Actions Reported as a Result of Profiles 



Action Taken 



Percent 

of 

Reporting 

PSROs 



Percent 

of 
Profiles 



Request for additional 

information/continued 

analysis 61.8% 29.4% 
Focus in (out) physicians, 

hospitals 46.3% 19.5% 
Modification of concurrent 

review 20.3% 5.3% 

MCE selection 18.7% 5.3% 

Special studies 5.7% 2.1% 

Objective setting 5.7% 2.0% 

Data quality projects 4.2% 1.3% 

Waiver withdrawn 2.1 % 0.5% 

Sanctions imposed 0.8% 0.1% 

Other* 39.8% 14.3% 

None 8.9% 3.2% 

None reported 47.4% 23.6% 



* Some examples of "Other" actions taken: 

"Educational sessions developed for hospital" 

"Report forwarded to State government agency" 

"Policies to be developed for pre-op LOS" 

"Hospital to initiate corrective action" 

"Criteria to be developed for appropriate 
outpatient procedures" 

"Information shared with HSA" 

NOTE: Since PSROs took more than one type of action 
as a result of the profiles, the percentage exceed 100. 



Impact. In this first assessment of profile analysis 
activity, it was difficult to assess the nature and level 
of impact that PSROs have been able to demonstrate 
as a result of their profiling activities. Of the 123 
PSROs reporting profile analysis, 31 (25%) reported 
achieving some type of impact as a result of at least 
one of their profiles. A total of 71 (8%) of the 911 
profile report forms indicated some impact. Table 81 
indicates the percentage of profiles in each "impact" 
category. Due to diversity of PSRO reporting and 
lack of documentation, this table should not be 
interpreted as representing actual impact, but rather 
what PSROs say they have accomplished. 



TABLE 81 
Reported Impacts of Profile Analysis 



Impact Achieved 



Percent of Profiles 
Reporting Impact (N=71] 



Decrease in review cost 
Decrease in LOS 
Decrease in pre-op LOS 
Decrease in inappropriate 

admissions 
Decrease in surgical rates 
"Other impacts" achieved 



19.8% 


23.9% 


2.8% 


4.2% 


1.4% 


47.9% 



100.0% 



"Other impacts" included such statements as: 
"decrease in percent of referrals" 
"decrease in days of care" 
"decrease in days certified at SNF level" 
"increase in denial activity" 
"improved use of physician advisor". 

6.5 Some Characteristics of PSROs that 
Use Data Effectively 

As part of its support of profile analysis, HSQB 
provided funds to a select group of seven advanced 
PSROs in an effort to enhance their profile analysis 
activity.^ This was followed by a contract which 
provided on-site technical assistance to those PSROs 
and an evaluation of the activities carried out under 
the special grants. In general, the contractor found 
that the PSROs used the grants effectively to extend 
their profiling capabilities and to perform studies 
of all aspects of care.^ This section of the PSRO 
Evaluation discusses the factors observed in these 
seven PSROs which were judged to facilitate or 
impede effective profile analysis.^ 

6.5.1 Recognition of Potential Uses 

PSROs that implement profile analysis effectively 
were judged to do so not because it is imposed 
upon them by law or regulation, but because they 
recognize that the analysis of patient care data is a 
tool to enhance their utilization review and quality 



'The seven PSROs were: Essex Peer Review Organization, 
Multnomah Foundation for Medical Care, National Capital 
Medical Foundation, New York County Health Services 
Review Organization, San Francisco PSRO, South Carolina 
Medical Care Foundation, and Southcentral Pennsylvania 
PSRO. 

' InterQual, Inc. "Final Report: PSRO Special Activity Projects 
in Profile Analysis," Report prepared under Contract HCFA 
500-78-0015, 1979. 

' "Effectiveness" or "success" in this context refers to the 
ability to use the data system to detect problems in care, 
to mobilize PSRO resources to diagnose and respond to 
these problems, and to monitor their resolution. 



135 



assurance objectives. PSROs that successfully conduct 
profile analysis have a clear understanding of how 
this activity can and should relate to other respon- 
bilities (e.g., concurrent review, focusing, data 
quality, objective setting). 

6.5.2 Commitment to Address Problems 

Another ingredient for the success of a PSRO's 
profile analysis seems to be the commitment of both 
PSRO physicians and staff. Section 7 of this report 
discusses the different orientations which may 
govern PSRO behavior; profile analysis was found to 
be most effective in those PSROs which are strongly 
goal-oriented and do not shy away from addressing 
problem areas. The PSROs that were the most 
successful had strong and committed staff and 
physicians who were willing to stand behind their 
analyses and pursue solutions. They investigated their 
data and developed profiles until they defined problems 
clearly. They then made those problems known and 
developed strategies to solve them. Some of the 
PSROs found that obtaining outside help to develop 
or critique their profiles was valuable. Such an 
independent opinion was viewed as more authoritative 
and objective and thus added forced to the profile 
findings. Even in these cases, however, it was ulti- 
mately the willingness of the physicians and staff to act 
on the results that led to productive results. 

Several of the PSROs mentioned the initial 
reluctance of their organizations to implement actions 
based on the profiles. This, coupled with resistance 
from targeted hospitals and physicians, made progress 
slow. It did seem that with each successive problem 
identified and attacked, the process became easier. 

6.5.3 Analysis and Implementation Capability: 

Organization of PSRO Staff and Physicians 

To be effective, a PSRO must adopt an organiza- 
tional structure which facilitates decision-making 
processes with regard to the results of the profile 
analysis activity, and facilitates communication at 
various levels within and outside of the PSRO. 
Communication is particularly important among the 
various operational units within the PSRO: among its 
staff, between the staff and physician decision-makers, 
and finally between both PSRO physicians and staff 
and area hospitals. 

Organization of Physician Committees. The 
organization of the PSRO's Board of Directors and 
physician committees is believed to be a key element 
in its successful performance of profile analysis. 
PSROs generally follow a model in which potential 
problems are identified by staff analysis, physician 
committees, review coordinators, physician advisors, 
or any one of a number of other sources. 
Analysis is then conducted by the staff with at least a 
minimal amount of physician input to ensure the 
clinical soundness of their recommendations and 
findings. The staff then presents its findings to a 
physician committee, usually called the Profile Analysis 



Committee, Data Committee, or Hospital Review 
Committee. It is at this point that a variety of different 
courses of action may be followed by the PSROs. In 
general, however, the course of action follows one 
of two directions. 

In the first decision-making model, the physician 
committee has a charge from the Board of Directors, 
usually using an explicit protocol, to seek more 
information, clarification or explanation of apparently 
aberrant patterns of practice from either the hospital 
or individual physician, and/or to work with the 
hospital to set objectives for the correction of those 
aberrant practice patterns which are not explained 
or justified. In these cases the Board of Directors 
retains its authority to levy sanctions such as 
rescision of delegation or intensification of concurrent 
review. The Board of Directors also retains 
responsibility to monitor the activities of this 
physician committee. 

In the second decision-making model, the physician 
committee takes no action but develops recom- 
mendations that go to the Board of Directors for 
its consideration before any action or external 
communication occurs. Since most problems 
identified using PSRO data require follow-up 
involving a request for more information, 
clarification, or explanations before any action 
can be taken, this model often takes more 
than one meeting of the Board of Directors before 
any information or communication is established 
with the hospital. Because the Board of Directors 
often does not meet as frequently as does the smaller 
committee, the profiling of an individual problem is 
drawn out over a long period of time, with resultant 
loss of interest and enthusiasm on the part of both 
staff and physicians. 

For the most part the first model was found to 
be more effective than the second. The reason seems 
to be that the Profile Analysis Committee or Hospital 
Review Committee is usually smaller in membership, 
meets more often, and is generally less political than 
the Board of Directors. The result is greater willingness 
and ability to engage in dialogue with the hospitals 
and physicians concerning areas identified as 
potential problems. 

Several of the PSROs studied had active and 
influential physician committees which led the profile 
analysis effort. In general, these committees were 
small, met frequently, and reported directly to the 
Board of Directors. Small committees are able to get 
work done and are usually more willing to make 
decisions and take action. By meeting often, the 
committee learns quickly and is able to keep abreast 
of rapidly changing issues. As the expertise of the 
committee increases, it becomes more effective at 
presenting profile results to the PSRO Board and its 
hospitals. Also, it does not lose continuity from 
meeting to meeting and, in cases where additional 
information is requested, it is able to reconstruct 
analyses from the previous meeting and move 
forward. By reporting directly to the PSRO Board or 



136 



its Executive Committee, the committee is close to 
the center of the power structure of the PSRO. 

Those committees that had been delegated the 
authority to take actions based on their findings were 
found to be more effective than those which had 
to pass their findings to other committees for action. 
While it is unlikely that a PSRO Board would delegate 
the authority to levy sanctions or rescind delegation, 
effective data committees are often free to develop 
and implement problem-solving strategies short 
of such serious steps, based on profile analysis. 

Organization and Qualifications of PSRO Staff. 
PSRO staffing patterns showed some general 
characteristics that seemed to relate to profiling 
success. Specifically, most of the PSROs studied had: 

• A staff specialist or expert in profile generation 
and analysis 

• Communication between data staff, review staff, 
and management 

• Support for the Executive Director for profiling 

The staff person who served as the leader for 
profile analysis was usually the Data Manager. 
Those most effective at profiling possessed a basic 
knowledge and understanding of: 

• The PSRO data system from data collection 
procedures through the structure of the data base 

• Medical data, especially coding systems and 
the meaning of various diagnoses and procedures 

• Computer programming and the use of software 
packages available 

• The general objectives, activities and needs 
of the PSRO 

Ordinarily, no one staff member is an expert in 
all four areas. The Data Manager may be an expert 
in two or three and rely on staff to add knowledge 
in the others. As long as the person doing profiling 
is aware of the issues in all of these areas, he or she 
can seek expert assistance as the need arises. Since 
the acquisition of this knowledge takes time and is 
rarely obtained through academic training alone, profile 
analysts with the most PSRO experience performed 
best. In addition, some clinical experience and 
knowledge and/or access to someone with this 
knowledge or experience, appeared to be important 
to effective profile analysis. Absence of clinical input 
often resulted in profiles that were logically and 
methodologically sound but clinically erroneous. 

Staff communication was seen to be as important 
as the individual skills of the Data Manager, 
and in some cases could compensate for the lack 
of expertise on the part of the Data Manager. In 
several of the PSROs where the profiling was 
imoacting directly on the hospital review process, staff 
communication was seen as a key to translating 
profile findings into review system modifications. 

Another benefit of staff communication was the 
addition of medical expertise to the profiling process. 
Several of the PSROs developed the necessary 



communication channels to provide both staff physician 
and nurse input into the early stages of the design 
and analysis of profiles. By establishing this 
cooperation at the outset, these PSROs were able to 
develop medically sound profiles before the profiles 
were presented to either physician committees or 
hospitals. In this way, they avoided having to 
argue their profile's medical validity and perhaps 
re-do them. 

In all seven of the PSROs studied, successful 
profiling was strongly encouraged and supported by 
the Executive Director. All Executive Directors 
interviewed were aware of the details of the profiles 
being developed and the analyses being made. In 
some cases, they designed and analyzed profiles 
themselves. Also, the Data Manager always had free 
access to the Executive Director even in cases where 
he did not report directly to him or her. 

6.5.4 Data Quality 

Effective profiles need to be based on complete 
and accurate data that reflect the actual processes 
that occurred in the hospital. Several of the PSROs 
studied have extensive monitoring programs to ensure 
that their data is complete, accurate, and timely. 
These programs include editing and periodic 
re-abstracting so that the quality of the coding in 
each hospital is known to the PSRO. No data quality 
control program can guarantee totally accurate 
coding, but a PSRO which is aware of the reliability 
of its diagnostic and procedural data can take this 
into account when it makes decisions based on 
the data. 

One of the PSROs studied was unable to make 
progress in profiling in part because its physician 
committees were not confident of the quality of the 
data they were analyzing. In other PSROs, physician 
committees were willing and able to deal with flawed 
data as long as they knew the extent of the 
inaccuracies. 

One way that PSROs deal with data problems in 
delegated hospitals is to remind a hospital of its 
responsibility as a delegated institution to provide 
timely and accurate data. PSROs have found that as 
they provide more data to hospitals and structure 
actions based on that data, the hospitals develop a 
greater interest in providing good data. 

6.5.5 Interactive and Flexible Data Systems 

In the history of each PSRO, the first data analyses 
are usually done using routine reports produced 
by the PSRO's data processing contractor. These are 
generally specified by contract, and rarely respond 
to the PSRO's immediate needs. As the PSRO 
develops, it may develop more elaborate analysis 
systems which are more closely adapted to its 
information requirements. Many PSROs find this 
method cumbersome, however. The most advanced 
PSROs have acquired the capability to analyze data 
interactively, using computer terminals in their own 



137 



offices and program packages specially adapted for 
health care data. 

Several of the seven PSROs which received special 
grants to further their profile activities already had 
interactive systems. The rest acquired them as a 
result of the grants. Acquisition of this capability 
had a dramatic effect both on the quantity and quality 
of the profiles produced. With only routine reports, 
the profile process was slow and labor-intensive. With 
the new capabilities, the organizations found 
that they were completing profiles and getting 
useful results at a greater pace. At times, as PSRO 
staff became more comfortable and adept at analyzing 
the data, useful profiles were being produced at a 
rate that exceeded the PSRO's ability to deal with 
the results. 

The PSROs identified several aspects of their 
flexible, interactive systems that made profiling easier 
and more productive: 

• Quick turn around to inquiries 

• Case-mix control 

• Ease of learning 

• Choice of displays and graphs 

PSROs found that quick turnaround time allowed 
them to follow a train of analysis when looking at a 
problem. They found that they could develop profiles 
and do a complete analysis in one session at the 
terminal. For example, if they were studying 
cholecystectomies, they could first look at all patients 
across all hospitals, and then look at the average 
length of stay within institutions and do further 
analysis to determine reasons for observed variations. 



The flexibility of the interactive system, coupled 
with its quickness, allowed the users to produce 
profiles that had medical significance, particularly by 
allowing the user to exclude or include specific types 
of patients. For example, if a local comparison was 
to be made against PAS norms from which deaths 
and long stays had been excluded, the user at the 
terminal was able to calculate the local length of stay 
after excluding those patients. Similarly, if during 
analysis a large variance was seen in overall patterns 
of length of stay, more homogeneous and meaningful 
patient groups could be isolated for further analysis. 
For example, in an analysis of cataract surgery 
rates, it is clearly important to analyze single cataracts 
and double cataracts separately. Some PSROs also 
excluded from the analysis any cataract patients who 
had diabetes, since the physicians felt that this 
additional condition might cause complications and, 
therefore, extensions in the length of stay. In this 
case, the analyst would be able to see if the presence 
of diabetes as a secondary diagnosis did in fact 
affect the length of stay. This ability to control 
case-mix, coupled with knowledge of case-mix issues, 
allowed the special activity projects to produce 
profiles that were medically meaningful to their 
physician committees. 

Users of the AUTOGRP system commented favorably 
on the ease with which the system could be learned. 
Users were able to develop an elementary under- 
standing of the system with only two or three days 
of training. They found that after several weeks of 
practice, they were able to do all but the most 
sophisticated analyses with the system. They also 
found that the graphical displays, statistical displays, 
and profile matrices were clean and understandable. 



138 



7. Organizational Effectiveness: Models 

and Measures for Assessing 

PSRO Effectiveness 



7.1 Introduction 

The PSRO is an organization established to control 
hospital utilization and to monitor the quality of care 
provided to Federal beneficiaries. In the preceding 
sections we have evaluated the effectiveness of the 
PSRO program with respect to control of hospital 
utilization and quality. In this section the perspective 
is changed slightly and the PSRO is studied as an 
organization. 

Over the past two years the PSRO evaluations 
have tried to explain the relative effectiveness of 
PSROs through case studies, ^ program site assess- 
ments, and project officer ratings.^ These efforts have 
made clear the need for the use of a theoretical model 
in assessing the organizational effectiveness of 
PSROs. 

This chapter presents the preliminary work done 
to identify those aspects of PSROs which affect their 
effectiveness. The chapter presents a review of 
organization literature, describes PSROs as organiza- 
tions, and finally presents approaches for future 
evaluation. 

7.1.1 Overview of Two General Models of Organizational 
Effectiveness 

Those who study the functioning of organizations 
have long been concerned with factors which 
account for differences in effectiveness among them. 
Although this is a common concern, there has 
developed, during the past two decades, divergent 
views as to how effectiveness is to be defined and, 
hence, what factors are most critical in determining it. 
There is currently a lack of consensus on what 
criteria are to be used to assess organizational 
effectiveness.^ This disagreement is due, in large 
part, to the varied theoretical models employed to 
conceptualize organizational functioning. Organiza- 
tional researchers are usually guided in their 
investigations by a theoretical perspective which 
contains specific assumptions about the nature of 
organizations and processes. The different theoretical 



' DHEW, PHS, HSA, OPEL, \/ol. 6: Case Studies of PSROs: 
A Contextual Analysis and Vol. 5: A Compretiensive Case 
Study: Tlie Colorado PSRO Experience. 

' DHEW, HCFA, Professional Standards Review Organization 
1978 Evaluation, pp. 230-282. 

^R. M. Steers, "Problems in the Measurement of Organization 
Effectiveness," Administrative Science Quarterly, 1975, 20, 
pp. 246-558. 

J. P. Campbell, "On the Nature of Organizational 
Effectiveness", pp. 13-55 and W. R. Scott, "Effectiveness 
of Organizational Effectiveness Studies", pp. 63-95 in 
P. S. Goodman and J. M. Pennings (eds) Weiv Perspectives 
on Organizational Effectiveness, San Francisco: Jossey-Bass, 
1977. 



perspectives suggest alternative measures of organi- 
zational effectiveness and alternative processes for 
accounting for differences in effectiveness. Since 
various researchers embrace differing perspectives, 
a given organization may be deemed effective by 
one researcher and ineffective by another. 

Two contrasting approaches to assessing organiza- 
tional effectiveness are outlined: the goal and the 
systems models. Each model is described briefly, 
its assumptions and limitations are presented, and 
some indicators of effectiveness consistent with 
each perspective are suggested. 

7.1.2. The Goal Model of Organizational Effectiveness 

The goal attainment or rational system model views 
organizations as instruments for the attainment of 
specified goals.' The goal approach is relatively 
straightforward and rests on the assumption that 
organizations exist to achieve a set of purposes or 
goals. Their effectiveness, then, is defined simply 
in terms of the degree of goal-achievement. The 
goats of the PSROs are often stated to be the 
reduction of unnecessary costs and the improvement 
of the quality of health care delivered to Federal 
patients in this country. Hence, a goal model of 
effectiveness would assess PSROs in terms of the 
extent to which they achieve these objectives. A 
general goal model has been used in all three of the 
PSRO evaluations. 

Several important assumptions are involved in the 
goal approach. First, it is assumed that organizations 
are purposeful, that they have goals or objectives 
toward which they are directed. Second, it is 
usually assumed that the organization's goals are 
set internally, that the organization is in the hands 
of rational decision-makers who have a more or 
less explicit set of goals toward the attainment of 
which the energies of the organization are directed. 
Third, it is assumed that the goals of the organization 
can be identified, that the explicit goals of the 
organization are the actual goals or, if not, that the 
latter can be determined. Fourth, it is assumed that 
these goals are relatively few in number, consistent, 
and sufficiently defined to be measured. Finally, the 
goal model assumes that the goals perform an 
orienting or directive function for members, 
determining the nature of their activities and specifying 
the criteria by which they are to be evaluated. ^ 

Many of these assumptions are being challenged 
by organizational researchers who question what 
appears to be an oversimplified conception of 
organizational goals. For example, Simon, with his 
focus on decision-making organizations, has argued 



* A. W. Gouldner, "Organizational Analysis," In R. K. Merton, 
L. Broom, and L. Cottrell, Jr. (eds). Sociology Today. New 
York: Basic Books, 1959. 

= L. B. Mohr, "The Concept of Organizational Goal," 
American Political Science Review, 1973, 67, 470-481; and 
S. M. Dornbusch and W. R. Scott, Evaluation and the 
Exercise of Authority. San Francisco: Jossey-Bass, 1975. 



139 



that organizations pursue not one or a few but 
multiple goals. He asserts: 

In the decision-making situations of real life, a 
course of action, to be acceptable, must satisfy a 
whole set of requirements or constraints. Some- 
times one of these requirements is singled out 
and referred to as the goal of the action. But the 
choice of one of the constraints, from many, is to a 
large extent arbitrary. For many purposes, it is 
more meaningful to refer to the whole set of 
requirements as the (complex) goal of the action." 

This more complex notion of goals is reinforced by 
the emerging view of organizations as comprised of 
collections of subgroups of participants who possess 
various differing social characteristics, are in 
different social locations, and who exhibit divergent 
views and interests regarding what the organization 
is and what it should be doing. In this view, 
organizational goals are determined by negotiations 
among shifting coalitions of participants whose 
influence in decision making reflects their relative 
power in the system.^ This conception of the 
organization as a political system can be expanded 
to include outside constituencies who hold goals 
"for" the organization « and who will attempt to see 
to it that their own interests are served by it. From 
this perspective, we should expect little common- 
ality in the criteria employed by the various parties 
who assess organizational effectiveness. This 
expectation is confirmed in a study of small businesses 
by Friedlander and Pickle, who report relatively low 
and sometimes negative correlations among effective- 
ness criteria imputed to owners, employees, creditors, 
suppliers, customers, governmental regulators, and 
citizens of the host community. They conclude that 
"organizations find it difficult to fulfill simultaneously 
the variety of demands made upon them". » 

The political model of organizations may be quite 
applicable to the analysis of PSRO effectiveness. 
Varied groups with possibly divergent interests relate 
to these organizations: national and regional officers of 
the program, local administrators, medical boards, 
hospitals, HSAs, etc. It is likely that such groups will 
differ on the goals served by the PSRO organizations 
and, hence, on the criteria which should govern 
the assessment of their effectiveness. 

Yet another development complicating the goal 
model of effectiveness involves recent challenges to 
the assumption that organizations necessarily exhibit 
a unified or consistent set of objectives. The political 



" H. A. Simon, "On the Concept of Organization Goal," 

Administrative Science Quarterly, 1964, 9, 7. 
' R. M. Cyert and J. G. March, A Beiiavioral Tfieory of Ttie 

Firm, Englewood Cliffs, NJ: Prentice-Hall, 1963; 

D. J. Hickson et al, "A Strategic Contingencies Theory of 

Intraorganizational Power," Administrative Science 

Quarterly, 1971, 16, pp. 216-229; and 

J. Pfeffer and G. R. Salancik, "Organization Decision 

Making as a Political Process," Administrative Science 

Quarterly. 1974, 19, pp. 135-151. 



model just reviewed allows for divergence and conflict 
among participants and constituencies but presumes 
their resolution through negotiation and power 
processes. In the end, the organization is assumed 
to pursue a single program. Alternative models now 
evolving suggest the utility of viewing some 
organizations as "organized anarchies" or as "loosely 
coupled" systems containing subunits that exhibit a 
high degree of autonomy and are capable of pursuing 
inconsistent objectives.^" This conception admits of 
the possibility that, with respect to any specific 
criterion of effectiveness, an organization might be 
both effective and ineffective depending on which 
component units are evaluated. These and related 
theoretical issues render the determination of criteria 
to be employed in examining organizational 
effectiveness a complex and controversial undertaking. 
While these amendments to and refinements of the 
goal model may seem to overwhelm it in complexities, 
this need not be the case. They may be viewed as 
useful reminders of certain "truths" which should 
not be overlooked in pursuing the goal model of 
effectiveness: 

1. Organizations are complex systems serving 
multiple goals. 

2. Different goals will be emphasized and/or 
preferred by different participants and 
constituencies. 

3. The selection of any particular goals as relevant 
for the assessment of organizational effectiveness 
is somewhat arbitrary, will be controversial, 

and will be supported by some and resisted by 
other organizational factions. 

7.1.3 The Systems Model of Organizational Effectiveness 

Numerous organizational analysts have criticized 
the goal model of organizational effectiveness and 
advocated the adoption of a systems perspective. It 
should be noted, however, that the two models are 
not inherently incompatible. Much of the dispute over 
which model should be adopted revolves around the 
degree of importance to be ascribed to the environ- 
mental context of organizations in explaining their 
effectiveness. 

In the systems persepctive, organizations are 
viewed as social units capable of achieving specified 
goals but concurrently involved with other activities 
required to ensure their survival as a system. 
Organizations are also viewed as highly interdependent 
with their environments and engaged in system- 



'J. D. Thompson, Organizations in Action. New York 

McGraw-Hill, 1967. 
' F. Friedlander and H. Pickle, " Components of Effectiveness 

in Small Organizations," Administrative Science Quarterly, 

1968, 13, pp. 289-304. 
'J. G. March and J. P. Olsen, Ambiguity and Ctioice in 

Organizations. Bergen: Universitetsforlaget, 1976; 

K. H. Weick, "Educational Organizations as Loosely 

Coupled Systems," Administrative Science Quarterly, 2976, 

21, pp. 1-19. 



140 



elaborating as well as system-maintaining activities. 
Finally, golas are not viewed as singular, static entities 
toward which organizations persistently strive, but as 
multiple dynamic targets which are adapted to suit 
changing social and political circumstances which 
impinge on the organization from its shifting internal 
composition and changing external environment. 

7.1.4 Distinctions Between Goal Attainment and 
Systems Models 

There are several important distinctions between 
the goal attainment and systems perspectives. First, 
organizational goals and their relation to the behavior 
of members seem much more problematic to the 
systems analyst than to the goal analyst. In general, 
the goal attainment model takes the organization's 
statements concerning its goals, its rules, and its 
norms much more at face value than does the systems 
model. The systems analysts note that there is 
frequently a discrepancy between the stated or the 
official goals of the organization and its actual 
objectives and between the roles specified for 
participants and their actual behavior. 

Second, the systems model does not focus on the 
set of narrowly defined goals as the unit of interest, 
but rather, the organization as a social unit capable 
of achieving certain goals. The organization is viewed 
as a self-maintaining system which must satisfy a 
stable set of internal needs at the same time that 
it must adapt to forces in the environment. Systems 
analysts insist on adding a set of support goals 
required to maintain the system to the output or product 
goals emphasized by the goal attainment model. 
Thus, organizations are viewed as multi-faceted 
systems that cannot devote all of their resources 
to the attainment of specific goals since some 
resources must be devoted to the maintenance of the 
system itself. 

Third, the systems model maintains that organizations 
do not find it possible to define a finite set of goals 
in any meaningful way, since the demands they face 
are dynamic and complex. Rather, organizations 
attempt to maintain their existence by avoiding 
behaviors which disturb internal affairs or their 
adjustment to the environment. This is really the 
most profound difference between the two perspectives. 
The main contention of the systems view is that 
organizations are more than instruments for attaining 
narrowly defined goals. Fundamentally, they are social 
groups attempting to adapt to and survive in their 
environments. Like other social groups, formal 
organizations are governed by the overriding goal of 
survival. The struggle for survival may persist even 
after an organization's goals have been successfully 
fulfilled. Indeed, the strain toward survival may 
sometimes foster the neglect or distortion of the 
organization's espoused goals." When survival is at 



Gouldner, op. cit., p. 405. 



stake, organizations will, if necessary, abandon the 
pursuit of their professed standards in order to save 
themselves. Or, under less severe constraints, 
organizations may modify their goals so as to achieve 
a more favorable position. It is because of such 
tendencies that systems analysts do not view 
organizations as means of achieving specific ends, but 
as ends in themselves. 

In general, the systems model is more complex 
than the goal attainment model in that both a larger 
number and a more diverse set of goals are added to 
the list of criteria of organizational effectiveness. 
Added to the output goals emphasized by the goal 
model is a set of support goals required in order to 
maintain the organization. To the extent that goals 
are used as criteria of organizational effectiveness, 
the above discussion of the goal attainment model 
will apply with certain corrections to include the 
effect of environmental relations on goal formulation 
and assessment. However, as mentioned earlier, 
output goals are not the only criteria used in assessing 
effectiveness from the systems perspective. 

The distinctions made between the two models are 
somewhat artificial and arbitrary: no real organization 
adheres to one model to the exclusion of the other. 
Nevertheless, the models are useful for analytical 
purposes. They are intended to highlight two aspects 
of these organizations, both present at least to some 
degree, but each somewhat contradictory to the 
other. To emphasize one aspect is to neglect the 
other; and to utilize both is to gain a more complete 
view of the functioning of these organizations. 



7.2 The Nature of PSROs 

Before examining the implications of the two 
effectiveness models for PSRO assessment, let us 
briefly attempt to describe the kind of organization 
being evaluated. PSROs are expected to function 
primarily as control systems. They do not themselves 
deliver medical services but are expected to monitor 
and control the performance of those who do. 
Control can be exercised in a number of ways. The 
PSRO can be described as a formalized, externally 
authorized and mandated local physician organization 
expected to function as a regulatory system exercising 
control via performance evaluations tied to financial 
and professional sanctions. The following will 
comment briefly on each aspect of this description. 

7.2.1 Formalized System 

PSROs are a formalized system of control. 
Formalized positions are those whose rights and 
responsibilities are explicitly defined and are available 
to all position occupants. The rights do not depend 
primarily on the personal characteristics of those who 
occupy the position; that is, they do not have to be 
"earned" or newly invented by each occupant. They 



141 



may, of course, be augmented or eroded by the 
performance and qualities of specific individuals; but 
the system is defined so as to minimize these 
personal factors. 

7.2.2 Externally Mandated System 

PSROs are required by law to operate in all 
geographic areas of the country. Physicians in local 
areas have some choice as to which organization 
is to be recognized as the PSRO, but they cannot 
choose whether such an organization is to be 
established or whether the organization, once 
established, will have the right to review records of 
their hospitalized patients whose care is supported 
by Federal funds. 

7.2.3 Externally Authorized System 

PSROs are empowered by Federal statute to review 
specified aspects of physician and hospital 
performance. The enabling legislation may be said 
to confer legitimacy on these systems by authoriza- 
tion. ^^ Authority has been defined as the normative 
regulation — both support for and circumscription of — 
power attempts. This regulation can occur from 
norms enforced by group processes among those 
subordinate to the power wielder — a type of authority 
system labeled as endorsement; or from norms 
enforced by processes among those individuals 
superordinate to the power wielder — a type of 
authority labeled as authorization. The authority of 
those empowered under the PSRO framework is 
authorized by legislative and administrative structures 
external to and superordinate over the professional 
performers whose performance is subject to regulation. 
This type of authority represents a departure from, 
and is expected to serve as an augmentation of, 
collegial authority. 

7.2.4 Regulatory System 

As noted, PSROs are not expected to deliver 
medical care but to evaluate and control the 
performances of those who do. Indeed, in the case 
of PSROs who operate "delegated" systems, they are 
twice-removed from health care delivery: they do 
not deliver medical care or directly review the 
performance of those who do, but "monitor" the 
systems within each local hospital that perform the 
direct review functions. Nevertheless, in a general 
sense, PSROs are expected to function as regulatory 
systems whose primary output is control. 

7.2.5 Local Physician Organization 

PSRO legislation protects the fundamental concept 
that physicians are the most appropriate individuals 
to evaluate the quality of medical services. 
Organizations designated as PSROs have been almost 
exclusively physician-sponsored, ^^ and the law 



requires that final medical decisions must be made 
by physicians. The requirement that physicians be 
directly involved in carrying out the control activities 
may be viewed as an attempt to gain the support of 
the physician community. By insisting that specified 
crucial roles in the organization (e.g., medical director, 
physician advisor) be filled by physicians, the 
organization is seeking to augment its legitimacy by 
securing endorsement to supplement its authorization. 
That is, the hope is that physicians' collegial norms 
will be harnessed in support of the new control 
system.i' 

In addition to its emphasis on physician control, 
the legislation embraces the assumption that peer 
review at the local level is the soundest method for 
assuring the appropriate use of health care resources 
and facilities. The presumption is made that whenever 
possible, control should be delegated to individual 
hospitals, and individual PSROs are granted 
considerable discretion in setting standards and 
establishing procedures of operation. 

7.2.6. Performance Evaluations 

Whereas accreditation agencies rely on structural 
measures such as the qualifications of staff or the 
elaborateness of the facilities in order to assess the 
adequacy of hospitals, PSROs concentrate on more 
direct measures of medical performance. Thus, a 
great deal of importance is placed on the development 
of monitoring systems which provide indicators of 
costs and/or quality based on what happens to 
patients — measures of care process or outcomes — as 
reflected in medical audits or patient-based abstracts. 

7.2.7 Financial and Professional Sanctions 

A variety of sanctions are available to PSROs to 
enforce their evaluations of the quality and 
appropriateness of care rendered. Primary among 
these is the disapproval of payment for care not 
judged to be medically necessary at the hospital 
level. Although rarely employed, the law provides 
a more extreme economic sanction, the suspension 
or revocation of the right to particpate in the Medicare 
and Medicaid programs. While the PSRO cannot 
directly affect a physician's right to practice, it can 
provide evidence to State licensing boards, regarding 
physician practice patterns. For present purposes, 
it is sufficient to note that the control system 
ultimately rests on real and palpable sanctions, and 
represents a departure from the kind of informal peer 
control that is traditional in medicine, in that it does 
not depend solely on moral or ethical pressures 
for its effect. 



'Dornbusch and Scott, op. cit. 

'Currently, four PSROs are sponsored by non-physician 
groups. 



" E. Freidson, "Speculations on the Social Psychology of 
Local PSRO Operations, "Proceedings: Conference on 
Self-Regulation. HEW Publication No. 77-621, June 1975, 
33-38. 



142 



7.2.8 Summary: The Nature of PSROs 

In their capacity as locally organized physician 
regulatory systems, the PSROs incorporate important 
features of earlier control systems governing the 
delivery of care in hospitals; however, insofar as they 
are more highly formalized, are externally validated 
and mandated, and have their evaluations backed by 
financial as well as professional sanctions, PSROs 
represent a new form of control over medical 
practice. 

Individual PSROs do not operate in isolation but 
as a part of a nation-wide system: in all, 195 PSRO 
areas have been designated to monitor the institutional 
delivery of health care in every area of the United 
States. A National Professional Standards Review 
Council of 11 physicians develops and distributes 
information to individual PSROs and reviews regional 
norms of care. In addition, in the 16 states containing 
more than three PSROs, a Statewide Professional 
Standards Review Council is to be created (about 
half are operational at the time of this writing). i= 

The PSRO system is not only designed to coordinate 
and standardize controls within health care institutions 
on an area-wide basis; it is intended to link this 
review process into a nation-wide system. These 
vertical linkages are intended to improve the capability 
of the area PSROs to monitor and regulate the cost 
and quality of health care and, at least in the long 
run, to influence the standards used by these agencies 
in carrying out their functions. 

It is important to emphasize that PSROs are, in 
their formal structure, "incomplete" organizations 
since their effectiveness is primarily dependent on 
the kinds of linkages they can establish with other 
organizations (i.e., hospitals, computer processing 
firms, fiscal intermediaries) rather than on the 
composition of their staff or the arrangement of their 
own "internal" structure. PSROs are connective 
organizations whose effectiveness must be assessed 
in terms of the links they forge between themselves 
and other systems. 

7.3 Observations and Postulates to be 
Tested about Factors Affecting the 
Effectiveness of PSROs 

While there has been no comprehensive study of 
all PSROs, the following discussion postulates factors 
that appear to need empirical testing. It is based on 
review of organizational and PSRO specific literature, 
PSRO program project officer ratings, PSRO site 
assessment reports, and on-site observations. 

7.3.1 From the Perspective of the Goal Model 

7.3.1.1 FACILITATING FACTORS 

If PSROs are viewed as mechanisms for controlling 
costs and assuring the quality of Federally funded 



medical care, a major difficulty confronting these 
organizations appears to be the lack of an as yet clearly 
efficacious technology for achieving these outcomes. 
Although this current evaluation and others indicate 
that PSROs are having modest effects on hospital 
utilization by Medicare beneficiaries and on compliance 
with the quality of care criteria, the efficacy of PSRO 
accomplishment of those outcomes is likely to remain 
in doubt. 

PSRO can be expected to have a positive impact 
on quality assurance and cost reduction activities 
in hospitals in two ways. First, the information 
gathering and processing activities of the PSRO make 
it, at least potentially, a force for change and reform. 
Gradually, data abstracting forms are becoming more 
uniform and systems are becoming more operational. 
The availability of comparable data sets from all 
care units will shed important light on the functioning 
of these units and will quickly expose "outliers" — 
organizations whose performance is markedly at 
variance from model standards of care. Second, PSROs 
can have some impact on the performance of these 
units by the types of training and support they 
provide for the review coordinators and, especially, 
the physician advisors, located in each care unit. 
Prior to the existence of PSROs, these physicians and 
quality assurance personnel operated entirely within 
the confines of their own hospitals. Often they were 
greatly limited in the types of data which they could 
collect or lacked information on the performance 
levels achieved in comparable facilities. Most 
important, they lacked any base of operations or form 
of support outside the facility in which they were 
located. The PSRO can provide an important support 
base and reference group for these physicians and 
medical personnel, and they can provide them with 
data which will allow them to evaluate the performance 
of their own facility in comparative perspective. 

7.3.1.2 LIMITING FACTORS 

As previously noted, the PSRO model requires the 
creation of an area-wide control system for monitoring 
physician and hospital performance. The theoretical 
components of a control system based on the 
evaluation of performance include: (1) clear allocation 
of evaluation and control responsibilities, (2) clear 
specification of evaluation criteria; (3) an inspection 
system for obtaining representative samples of 
performances; (4) an appraisal system for comparing 
sampled performances with evaluation criteria; 
and (5) a feedback system for communicating 
evaluation results and sanctioning decisions based on 
these results back to performers. ^^ The survey of the 
relevant literature and exploratory observations 
suggest that there are difficulties with all of these 
system components as they presently operate in 
PSROs. Problems include the delegation of evaluative 



'^See Section 8.2.7. 



" Dornbusch and Scott, op. cit. 



143 



functions to hospitals, some of wliicli are not capable 
of carrying out these responsibilities objectively and 
efficiently; a lack of consensus on appropriate 
criteria for evaluating medical care performance; an 
absence of clear criteria for selecting performance 
units to be sampled; a lack of uniformity in the ap- 
praisal of performance across units in the same 
PSROs; and the absence of timely and pertinent 
feedback from the PSRO to performers. 

Although, from one point of view, all of these 
deficiencies are technical problems and presumably 
have technical solutions, in a larger sense they are 
not simply technical in nature. Thus, while it is true 
that more knowledge and more experimentation 
will produce improvements in criteria — the selection 
of salient properties, the esta'oii'siment of specific 
and realistic standards — it is also important to note 
that criteria setting is hampered by the political 
issue of whether PSROs are to develop and enforce 
a national set of standards or whether allowances 
are to be made for regional and local differences in 
medical customs and practices. This is not a 
technical but a political decision that cannot be 
resolved by advances in technical knowledge. 

Similarly, issues concerning what proportion of 
cases are to be reviewed by PSROs — the size of the 
sample of medical performances to be drawn — is not 
simply a question which is based on information on 
population size and amount of variance exhibited; 
it is also importantly affected by the amount of funds 
allocated for data collection and processing. The 
recent decision regarding the focusing of medical 
review procedures is driven more by an interest in 
reducing expenditures within the program than by 
a recognition that 100 percent samples are 
not required for review and monitoring effectiveness. 
As a technical matter, focusing can only safely 
occur after sufficient baseline data are available to 
permit informed judgments concerning what areas of 
practice are clearly within acceptable guidelines. 

Appraisal functions also face technical and practical 
problems. Currently PSROs are generally expected 
to delegate these functions to hospital UR 
committees which are themselves highly variable in 
effectiveness. In addition to the variation in the 
extent of personnel training provided by PSROs, the 
hospital context in which they function may be 
expected to provide variable supports for and 
constraints on the nature of their reviews. Physician 
advisors are expected to adapt to practice differences 
within hospitals to a greater extent than they 
introduce area-wide expertise and/or standards. 

Feedback mechanisms that bring results of 
evaluation efforts to the attention of practicing 
physicians are a particularly important but still weak 
area of PSRO functioning. Reviewers who challenge 
admissions or additional days of stay rarely address 
these challenges directly to practicing physicians. 
Review coordinators raise issues with physician 
advisors, and only if the advisor concurs is the 
individual physician contacted. As a consequence, 



most physicians have no regular contacts with the 
review system. MCE and profile studies which permit 
broader and comparative assessments of medical 
care practice currently suffer from delays in the 
gathering and processing of information such that 
when feedback to hospitals or physicians does occur, 
it is likely to be considered untimely. While some of 
these deficiencies will undoubtedly improve over 
time, others will require more than time to remedy. 
The PSROs are armed with a variety of sanctions 
with which to enforce compliance to (or exact a 
penalty for noncompliance with) their decisions. It 
is noted, however, that all of the sanctions involved 
are negative: positive incentives are not built into the 
operation of the program to reward effective or 
successful performance. The only contacts which 
occur between practicing physicians and PSRO 
representatives are those occasioned when a 
physician's judgment is being questioned. Physicians 
and/or hospitals are not sought out in order to be 
commended for their effective efforts. None of the 
cost savings which are presumably produced by 
actions of the PSRO and its cooperating units are 
passed on to the performing units to reinforce 
their continued cooperation. 

7.3.2 From the Perspective of the Systems Model 

If PSROs are viewed as organizations concerned 
with their own survival, it is clear that PSROs are 
weak and highly vulnerable organizations. The 
general public is unaware of PSRO existence. 
Although PSROs are regarded as "physicians' 
organizations", many physicians, at best, tolerate 
the existence of PSROs as an alternative to some 
less desirable form of control and, at worst, are 
actively hostile to this new form of "bureaucratic 
meddling." 

Following a development phase of growth and 
increasing appropriations, PSRO funding has leveled 
off in recent years, as might be expected, but the 
most recent budget year shows an absolute decline 
in funding level from $151 to $144 million. Given the 
current rate of inflation, this decline is even larger 
than it appears. At a time when many individual 
PSROs have recently moved from planning to 
conditional status, this decline in support must appear 
as a serious threat on the horizon. 

In addition to the serious problems of lack of 
legitimacy among physicians and instability of the 
funding future, PSROs lack control over some of the 
critical aspects of their own functioning. Costs of 
review are not covered directly by Federal grants to 
PSROs. In most cases, review functions are delegated 
to hospitals. Each hospital is expected to negotiate 
an arrangement with PSRO to pay the costs of 
delegated review, and although Medicare fiscal 
intermediaries are required to request PSRO review 
of and comment on all costs claimed by delegated 
hospitals which would result in the hospital being 
reimbursed at a higher unit cost rate than was 



144 



negotiated, final settlement authority rests with the 
intermediaries. The lack of administrative and/or 
financial control in this area may pose operational 
problems for PSROs. 

Finally, PSROs may also be said to suffer from 
"the liability of newness." " Newly established 
organizations, and particularly new forms of 
organizations, have a difficult time; most do not survive 
for very long. The process of creating new roles, new 
skills, and new incentive structures so as to get 
maximum performance has high costs in time and 
temporary inefficiency. New organizations must 
develop new connections with existing roles and 
organizations, and particularly in a "crowded" sector 
such as health care, may be expected to create 
suspicion and conflict as existing organizational and 
occupation groups "size up" the newcomer. 

7.3.3 Implications 

The foregoing description suggests that PSROs 
are likely to be weak organizations. They lack a strong 
technically based evaluation system which enables 
them to readily and regularly achieve their avowed 
objectives, and they lack a strong political foundation, 
both at the national and the local levels, which can 
ensure their continued survival. Moreover, in a very 
real sense, effectiveness in terms of goal attainment 
criteria will only serve to undermine their effectiveness 
in terms of systems criteria. That is, the more 
rigorous and active PSROs are in evaluating and 
sanctioning doctors and hospitals, the more likely 
they are to undermine their already shaky base of 
support, threatening their survival. Organizations 
confronting this type of dilemma may be expected 
to "decouple" their structures and activities in an 
attempt to satisfy the conflicting demands.^* 

More specifically, we might expect to see PSROs 
exerting much effort to conform to the formal 
requirements imposed on them by the regional and 
national offices. These requirements include 
conforming to report deadlines, working within budget 
constraints, developing appropriate interconnections 
with hospitals. State agencies, fiscal intermediaries, 
and introducing new programs as required by Federal 



" Stinchcombe, Arthur, "Social Stucture and Organizations," 
in James G. March (ed.), Handbook of Organizations. 
Chicago: Rand McNally, 1965. 

" Weick, K. H., "Educational Organizations as Loosely 
Coupled Systems," Administrative Science Quarterly, 1976, 
21, pp. 1-19. 

M. Davis, T. E. Deal, J. W. Meyer, B. Rowan, W. R. Scott, 
and Anne Stackhouse, The Structure of Educational 
Systems: Explorations in ttie Tfieory of Loosely-Coupled 
Organizations. Stanford, CA: Stanford Center for Research 
and Development in Teaching, June 1977; J. W. Meyer, 
W. R. Scott and T. E. Deal "Institutional and Technical 
Sources of Organizational Structure," Paper presented at 
a conference on Human Service Organizations, Center of 
Advanced Study in the Behavioral Sciences, Stanford, 
California, March 3, 1979. 



initiatives. However, at the operational level, the 
PSRO may not implement its control activities with 
vigor. Local agencies will not want to alienate hospital 
administrators and physicians, and are likely to 
attempt to couch their demands in ways that will 
make them acceptable to "host" organizations. 
They may choose to work primarily with a small set 
of "cooperative" physicians — typically who are 
already involved in UR committee work — as well as 
with existing hospital and physician quality assurance 
mechanisms. In the typical case, review activities 
are "delegated" to individual hospitals. These 
hospitals, to a large extent, determine their own UR 
processes, decide what MCEs to carry out, etc. They 
provide a clear instance of a "loosely coupled" 
system, in which the individual hospitals have 
considerable autonomy in carrying out review 
processes and in determining what standards are 
to govern their activities. In some respects, then, 
the PSRO agencies function as a "buffering" system 
between the Federal program and its requirements 
and the local organizations and their vested interests 
and customary ways of operating. If it is to survive, 
the PSRO agency must conform to the requirements 
of the Federal agency, but must not alienate the 
local organizations on whom it is dependent. 

Another basis for expecting PSRO systems to be 
"decoupled" is provided by the fact that a fairly 
rigid organizational structure has been imposed on 
these systems from the outset. In order to be funded, 
PSROs must conform rather closely to a set of 
structural guidelines which indicate what positions 
are to be established and the appropriate reporting 
and authority relations among them. Rather than 
permitting the structure to vary depending on the 
specific medical environment to which it must relate, 
structures are expected to conform to a single 
pattern. The variability of the environments forces 
PSROs to adapt their actual patterns of activities 
to local conditions. Structures remain uniform while 
activities vary: a situation virtually forcing a 
decoupling of structures from activities. 

More generally, it appears that the PSRO may be 
usefully viewed as a type of "institutionalized 
organization." This model was originally developed 
by Meyer and Rowan i" to describe the distinctive 
structural features of educational organizations; 
however, the model has been expanded to include 
other types of public service organizations. 

Institutional organizations do not arise as structures 
to coordinate and regulate technical work processes 
but come into existence in conformity to and as a 
reflection of institutional rules and codes. Their 
survival is not based on effectiveness and efficiency 
of market transactions but rather on conformity to 
externally defined rules. The actual work performed in 
these organizations is not tightly regulated by the 



' J. W. Meyer and B. Rowan, "Institutionalized Organizations: 
Formal Structure as Myth and Ceremony," American 
Journal of Sociology, 83 (Sept. 1977), pp. 340-363. 



145 



administrative system: rather it tends to be delegated 
to professionalized or certified workers. These 
participants exercise a great deal of discretion in 
their conduct of work and are not subjected to close 
scrutiny. When evaluations do occur, the emphasis 
is placed on conformity to rules and procedures 
(processes) rather than on outcomes achieved. "Thus, 
the technical organization faces in toward its 
technical core and turns its back toward the 
environment; the institutional organization turns its 
back on its technical core in order to concentrate 
on conformity to its institutional environment." ^o 
To the extent that the pursuit of its official goals 
undermines the survival capabilty of the PSRO, one 
may expect these organizations to deemphasize the 
implementation of these goals while meeting the 
reporting requirements of HCFA. 



7.4 Applicability of the Goal and Systems 
Models for Assessing PSRO 
Effectiveness 

7.4.1 The Goal Model 

The goal model focuses attention on the goals of 
the organization as a basis for assessing their 
effectiveness: the greater the degree to which an 
organization achieves its goals, the greater its 
effectiveness. While it would appear to be a straight- 
forward matter to assess the effectiveness of any 
given organization using this type of criterion, as 
with many organizations, there is an immediate problem 
in assessing the effectiveness of PSROs. The 
announced goals to be pursued by this agency 
are defined as minimizing the costs of providing care 
to Federally supported patients and assuring that 
professional standards are observed in the provision 
of care. As general goals, reducing costs and 
maintaining quality are not inevitably in conflict, but 
they do press in opposite directions. The general 
goals to be pursued by PSROs are sufficiently 
ambiguous as to be subject to varied interpretation, 
and "honest men" may disagree about the weights 
to be assigned to their two principal concerns. 

Given that the question of goals can be resolved — 
if not in general, then sufficiently to permit the 
delineation of locally acceptable goals — what are 
the processes that must be performed in order for 
a PSRO to be effective? If the above described 
view of these organizations is correct, then they 
need to be able to arrive at valid and reliable 
performance evaluations of the units — both hospitals 
and physicians — whose work is to be regulated. A 
modified version of the Dornbusch and Scott general 
model for analyzing control systems based on 
performance evaluations will be used in analysis of 
the effectiveness of PSROs as goal attainment systems. 



7.4.1.1 CONTROL SYSTEM COMPONENTS 

Dornbusch and Scott identify four components 
of a control system based on performance evaluations. 
The four components are: (1) mechanisms for 
allocating tasks, (2) mechanisms for determining the 
criteria of evaluations, (3) mechanisms for selecting 
the sample of work to be evaluated, and (4) 
mechanisms for appraising the work sample in order 
to arrive at a performance evaluation. A fifth 
component appropriate for PSROs is mechanisms for 
providing feedback of information to the performers, 
including the threat of sanction if performance remains 
unsatisfactory. Each of these components is briefly 
described and their applicability to PSROs discussed 
below. 

1. Task Allocation 

Task allocation is defined as the act which 
determines who is to perform a given task and, 
hence, specifies who is to be evaluated for it^^ As 
noted above, the major tasks performed by PSROs 
are control or monitoring tasks. A crucial decision 
made by PSROs is whether or not to delegate review 
tasks to participating units such as hospitals or to 
perform these tasks directly. Currently, most PSROs 
delegate review functions to delivery system 
personnel, retaining for themselves the more 
general — and remote — tasks of training the review 
personnel and monitoring their performance. 

2. Criteria Setting 

If performance evaluations are to be made, 
criteria must be established on the basis of which 
the performances can be judged. It is useful to 
distinguish three elements present in any set of 
evaluation criteria: (1) dimensions or properties of 
the performance which are singled out for attention, 
e.g., the cost of care or its quality (of course, many 
more specific sub-properties can be identified as 
comprising each of these more general dimensions); 
(2) if more than one property is at issue, weights must 
be assigned to specify the relative importance to 
be placed on each; and (3) standards must be 
established which specify desirable levels of 
performance to which actual performances can 
be compared. 22 

Clearly, the setting of criteria is one of the most 
crucial components of any control system. It is 
recognized to be a particularly difficult and hazardous 
undertaking when the performances to be assessed 
are highly complex, as is the case with medical 
care. Present practice apparently leaves much to be 
desired. The major conclusion of the Institute of 
Medicine committee, which reviewed a sample of the 



Meyer, Scott and Deal, op. cit. 



-^Dornbusch and Scott, op. cit., p. 136. 
-Dornbusch and Scott, op. cit., pp. 138-139. 



146 



"better quality assurance programs" just at the 
beginning fo the PSRO period, was pessimistic: 
The stated goals are to ensure high quality 
medical care at a reasonable cost. But the goals 
are not expressed in terms that permit measurement 
of the degree to which they are achieved. The 
margin by which quality might be improved is not 
known. Even rough estimates of the magnitude of 
currently inappropriate care were unavailable.^^ 

This discussion does not present a detailed 
examination of criteria setting activities in PSROs, 
but only comments on the three elements identified. 

(a) Determining what dimensions or properties of 
performance are to be singled out for attention 

is a difficult task. Patients are highly variable In the 
services they require, and hospital systems and 
physicians also vary in the types of services they are 
equipped to provide. Most approaches to quality 
assessment begin by differentiating patients by 
diagnosis categories and then specifying relevant 
properties within these categories. Although there 
is now fairly widespread agreement on a patient 
classification system, there is much less agreement 
on what properties of patients or performances are 
identified as salient for purposes of evaluation. 
Innovation by hospitals and PSROs is encouraged 
in the area of quality assessment. Similar complexities 
are encountered in performing UR activities, although 
there is currently widespread use of certain crude 
measures of extensiveness of use (discharge rates 
per eligible beneficiary) and duration of care 
(average length of stay and total days of care per 
eligible beneficiary). These measures emphasize 
duration and ignore differences in service intensity, 
and typically they are not adjusted for differences 
In patient mix. 

(b) Little explicit attention is devoted to determining 
weights to be assigned when more than one property 

is identified for use in evaluating performances. 
Partly because this aspect of criteria setting is 
neglected, properties receive unknown weighting, 
not by choice but by default. 

(c) Some discussions of PSRO criteria distinguish 
between norms and standards. Norms are defined 

as documented measures of "usual" (mean or median) 
performance while standards express the range of 
acceptable variation from norms in assessing 
performance. 2* In early stages of the program, since 
little data on performances exist, there is a tendency 
to adopt average levels of care (within a region) 
as the normative standard and seek to reduce 



^ Institute of Medicine. Assessing Quality in Health Care: 
An Evaluation. Washington, D.C.: National Academy of 
Sciences, 1976, p. 3. For further discussion of estimated 
levels of inappropriate care, see Gertman, op. cit. 

^ Health Policy Program. Cooperation Between Health 
Systems Agencies and Professional Standards Review 
Organizations. San Francisco: University of California, 
School of Medicine, Health Policy Program. Jan. 31, 1977, 
p. 22. 



variations around that mean. However, it is widely 
known that such norms vary greatly between regions 
of the country and among PSRO areas for reason 
that are not well understood.^^ 

3. Selecting the Sample of Work to be Evaluated 

Two basic types of sampling decisions must be 
made. First, evaluators must choose what indicators 
they are going to employ in assessing the actual 
work performed. Three classes of indicators are 
typically differentiated: structures, processes, and 
outcomes.2o Most of the indicators employed in 
medical care evaluations utilize process measures. 

The second decision involves what proportion 
of the relevant units are to be examined. Few 
performance assessment systems attempt to obtain 100 
percent samples of the work performed. Most must 
settle for smaller samples. It is often the case that 
work samples are poorly drawn, are severely biased, 
or nonrepresentative of the large population they 
are designed to reflect. Because of the highly 
differentiated nature of the patient population (or 
the types of performances), work samples are often 
plagued by small numbers of similar cases. 

When we turn to more specific UR functions and 
MCE studies, it is not to easy to ascertain what 
sampling decisions are made. Some hospitals began 
to conduct their concurrent reviews with 100 percent 
samples of Federally financed patients, but this 
practice has proved to be too expensive and, in 
any case, not cost effective so that it largely has 
been abandoned in favor of "focused reviews." 
Unfortunately, the basis on which focusing takes 
place — that is, the criteria to be employed in the 
sampling decision — are neither made explicit nor 
applied uniformly by all hospitals within PSROs. 
Similarly, MCE studies vary tremendously by topic and 
within PSROs.2^ 

Returning to the routine UR functions, it is important 
to stress that sampling decisions are made at several 
levels: the selection of cases to be routinely screened 
by the review coordinators, the selection of cases 
to be passed on to the physician reviewer, and the 
selection of cases to send on to the physician review 
committee. Each decision level relates to a different set 
of cases produced by the sampling screens operating 
at lower levels in the hierarchy. 



^R. Deacon, J. Lubitz, M. Gornick, and M. Newton, "Analysis 
of Variations in Hospital Use by Medicare Patients in 
PSRO Areas, 1974-77", Health Care Financing Review, 
1 (Summer 1979), pp. 7&-107. 

^A. Donabedian, "Evaluating the Quality of Medical Care," 
Milbank Memorial Fund Quarterly 46 (July 1966), Part 2, 
166-206; 

E. A. Suchman, Evaluative Research. New York: Russell 
Sage Foundation, 1976; 

W. R. Scott, "Effectiveness of Studies of Organizational 
Effectiveness," in P. Goodman and H. Pennings (eds), 
New Perspectives on Organizational Effectiveness. San 
Francisco: Jossey-Bass, 1977, 63-95. 

""See Section 5. 



147 



Some types of data are routinely gathered on all 
patients, and a uniform reporting form (PHDDS) has 
been created to guide the gathering and abstraction 
of data from patient records for all eligible patients. 
This information is computerized, providing a 
continually up-dated data bank on the experience 
of patients in each care unit under the jurisdiction of 
the PSRO. These data can be used for profile 
analysis — to make comparisons of hospital 
performance over time, comparisons between 
hospitals on specified diagnostic categories, and 
even comparisons between different physician 
providers. However, at the current time, many problems 
plague the development of these systems including 
the use of varied record systems within hospitals, 
delays in getting abstracted data into the computer, 
low priorities assigned by computing firms to PSRO 
data processing, and inexperience in handling such 
data sets at the PSRO level. Because of such problems, 
data collected do not often reflect the current 
performance of the units under review. 

4. Performance Appraisal 

Appraisal functions consist of comparing sampled 
performance values against standards contained in 
the criteria to arrive at a performance evaluation. 
There is often considerable discretion involved in 
the appraisal function, especially in cases where 
criteria are not very explicitly defined and performance 
values are not readily measured. Obviously, such 
functions occur at several levels in the PSRO structure. 
For UR, appraisals are made by review coordinators, 
medical advisors, and medical review boards, the 
first level applying general rules routinely to specific 
cases, the middle level handling ambiguous cases 
or those which appear not to to meet the established 
criteria; and the third level handling appeals by 
physicians whose cases have been turned down by 
medical advisors. 

For MCE functions, UR committees in hospitals 
perform appraisal functions instead of or in addition 
to the reviews carried on by appropriate committees 
at the PSRO level. 

5. Feedback 

In an ideally functioning regulatory system, salient 
performance values would be regularly sampled, 
compared to appropriate standards, and discrepancies 
between desired and actual performances detected. 
Such discrepancies or apparent failures would then 
be called to the attention of the responsible 
participants. Corrective procedures would be called 
for; sometimes, warnings would be issued; and, if 
necessary, reimbursement would be denied. In this 
manner, in theory, regulatory systems attempt to 
impact future performance by sanctioning past 
performance. 

It is also important to note that the mere existence 
of a control system which regularly and routinely 
gathers information on physician and hospital 
performance may be expected to exert some effects 



on performance, even in the absence of explicit and 
"specific feedback from controllers to performers. 
Performers' awareness of the existence of monitoring 
systems can exert constraints on their behavior. This 
is the type of effect produced by a patrol car on 
the speed of freeway traffic in its vicinity. 

The actual effect of the PSRO system on physician 
and hospital performance is expected to be 
attenuated due to the several factors which 
represent departures from the idealized model just 
sketched. First, the use of any of the more extreme 
formal sanctions available to PSROs is extremely rare. 
Indeed, in general, only a small proportion of 
decisions concerning admission or length of stay are 
challenged by review coordinators, and an even 
smaller number are disallowed by the physician 
advisors.28 Also, it is important to note that the 
typical physician rarely has any contact at all with 
the PSRO review process; monitoring is based on 
chart review and patient abstracts, and if a physician's 
decision is questioned by the review coordinator, the 
question is raised with the physician advisor, not 
the practicing physician. It is only if the advisor 
questions the decision that the responsible physician 
is confronted. Thus, the typical physician has no 
regular contact with the PSRO review mechanism in 
its day-to-day operation. 

7.4.1.2 EFFECTIVENESS INDICATORS 

The goal model takes PSROs "at their word"; it 
assumes that PSROs were created to perform certain 
regulative functions and are to be evaluated in terms 
of their effectiveness (and efficiency) in obtaining 
these objectives. The model presumes that PSRO 
effectiveness is primarily determined by their ability 
to manage certain technical problems — ^for example: 
their ability to develop specific criteria for evaluating 
the cost and quality of medical performance; their 
ability to select a representative sample of relevant 
performance units to review and to establish an on-line 
data system which will enable routinized inspection 
of certain characteristics of all discharged patients; 
their ability to appraise performance by comparing 
selected performance measures with appropriate 
standards; and their ability to *eed back information 
based on these reviews so that future performances 
mav be improved by knowledge of past deficiencies. 

Thus, some types of indicators suggested by the 
goal effectiveness model to assess the effectiveness 
of PSROs would include: 

1 . Task Allocation 

• The ability of PSROs to select, recruit, train, 
and influence review coordinators and physician 
advisors in the member hospitals; 

• their ability to effectively communicate with and 
exert influence on hospital UR committees; 



'See USDHEW, HCFA, Professional Standards Review 
Organizations: 1978 Program Evaluation. 



148 



• their ability to supply technical assistance to 
hospitals in carrying out and following through 
on medical care evaluations. 

2. Criteria Setting 

• The use of clear and specific criteria in assessing 
the costs and quality of medical care; 

• the dissemination of information concerning 
norms, standards, etc. to all participating 
hospitals; 

• the uniformity of application of standards across 
all participating hospitals; bases of exceptions 
should be clearly detailed; 

• the establishment of narrower rather than 
broader ranges of acceptable variation around 
standards; 

• the establishment of norms for patient and/or 
diagnostic categories that are defined more 
narrowly rather than more inclusively; 

• the establishment of norms based on more recent 
rather than on more remote data concerning 
past performance; 

• the active involvement of local physicians in 
setting standards and reviewing criteria. 

3. Sampling 

• The proportion of Federal admissions under 
review in each participating hospital; 

• the extent to which hospitals in the area employ 
uniform data forms and abstracting procedures; 

• the adequacy of justification for selection of 
MCE studies by participating hospitals; 

• the adequacy of justification for "focusing" of 
reviews by participating hospitals; 

• the currentness of abstract-based data set 
available to PSROs for analyses. 

4. Appraising 

• The extent of training of review coordinators 
provided by PSROs; 

• the extent to which PSROs select, train, and 
influence physician advisors. 

5. Feedbacl< 

• The extent to which PSROs provide timely and 
specific feedback to participating hospitals 
concerning performance deficiencies; 

• the extent to which a hospital's UR committee 
receives training and technical assistance from 
PSROs in interpreting their own and PSRO- 
supplied data. 

6. General 

• The number of multi-hospital and/or area-wide 
MCE studies conducted during specified period; 

• the proportion of MCE re-audit studies conducted, 
given an indication of unsatisfactory condition 
during first audit, during specified period; 

• the extent to which there occurs reduction in 
variance of performance among hospitals in area. 



7.4.2 The Systems Model 

The systems model treats the organization as an 
adaptive system — a system primarily concerned 
with its own survival. Although the organization may 
define explicit output goals and devote energy to their 
attainment, as emphasized by the goal model, the 
systems model assumes that organizations attempt 
to maintain their existence by avoiding behaviors 
which disturb internal affairs and agreements or their 
adjustment to the environment. If and when the 
pursuit of output goals conflicts with the organization's 
survival or "support" goals, the organization is 
expected to abandon or compromise the former for 
the sake of the latter. 

Whereas the goal model of organizations produces 
a set of effectiveness criteria which emphasizes 
the organization's ability to function as a technical 
system, the systems model stresses the ability of 
the organization to generate sufficient support and 
legitimacy to survive as a political system. 

7.4.2.1 SYSTEMS CONSTITUENCIES 

Assuming a systems perspective, one must identitfy 
the central actors and constituencies, both within 
and outside the organization, which provide the 
major resources that sustain the organization. A 
cursory inspection of PSROs suggests the following 
central participants or interests, by level (partial, 
not exhaustive listing): 

1. National 

U. S. Congress; relevant committees and sub- 
committees which oversee health policy. 

Department of Health, Education, and Welfare 
(DHEW). 

Health Standards and Quality Bureau (HSQB): 
operational agency. 

Health Care Financing Administration (HCFA); 
planning and evaluation units. 

American Medical Association (AMA). 

American Association of Professional Standards 
Review Organizations (AAPSRO). 

2. Local 

PSRO regional office. 

Affiliated hospitals and extended care facilities; 

UR committees. 
Medicare fiscal intermediaries (Blue Cross, 

insurance carriers). 
State Departments of Health and Social Services. 
Local Health Systems Agency (HSA). 
Data processor. 
Abstracting services. 
County medical societies. 

3. Internal 

Executive director and assistants. 
Medical Director. 
Physician board; committees. 
Physician membership. 



149 



Medical Advisors. 
Data manager. 
Review Coordinators. 

Even this abbreviated listing illustrates the 
complexity of the organizational environment within 
which PSROs operate. A few comments about the 
relations among the levels (which are themselves 
somewhat arbitrary in definition): It is important to 
note that survival of the organization is an issue at 
both the national and local levels. At the local level, 
an individual PSRO may lose its contract with the 
government or its legitimacy with local physicians and 
so go out of existence; this is a rare event, but it 
has happened in past years and will undoubtedly 
continue to happen. At the national level, there is the 
possibility that the entire program will be eliminated 
perhaps because it is believed to be ineffectual or 
becomes too controversial or is replaced by an 
alternative program. There is a kind of reciprocal 
interdependence between the national and the local 
programs: the persistence of both levels of activity is 
dependent on generating political support and 
legitimacy and refraining from alienating or upsetting 
important constituencies. 

There are also important connections between 
local and national levels in a more direct sense. The 
same types of interests are frequently organized at 
both a national and local level, and we would expect 
to see coalitions formed and joint actions involving 
groups across levels. For example, the American 
Medical Association might take a stand on some 
issue involving PSRO functioning which position in 
turn may be embraced and advanced by local county 
medical societies. Alternatively, executive officers of 
PSROs have formed their own professional association; 
AAPSRO, to protect their interests at the national 
level. Similarly, many of the other organizational 
participants (e.g., Blue Cross, Health Systems 
Agencies) are organized at both local (regional) 
and national levels. Given these "nested" relations, 
interactions among actors at any level become 
exceedingly complex since all actions tend to entail 
both a horizontal and a vertical dimension. Actors 
are not free to adapt completely to local circumstances 
in relating to one another but are restricted by 
hierarchical constraints. For example, PSROs and 
health maintenance organizations are highly con- 
strained both by their respective policies and proce- 
dural definitions, but also by the realization that 
innovations or modifications negotiated between local 
affiliates may create precedents which are viewed as 
undesirable by their respective hierarchies. 
Accommodation, resource acquisition and, generally, 
survival within these complex networks of relations 
is no simple matter and may consume much of the 
energy of the organization. 

Examination of the constituencies related to PSROs 
is likely to reveal the presence of strong conflicting 
demands between national (centralized) and local 
(decentralized) interests. In a very real sense these 
conflicts are built into the very structure of the 



organization which, as noted at the beginning of this 
section, consists of Federally funded, centrally 
organized organizations of local physicians. Since 
the early 1960s, there have been a number of 
decentralized Federal programs (e.g., community 
action agencies, community mental health centers, 
model cities) which are Federally funded and guided 
by generalized Federal policies, but locally admin- 
istered in order to accommodate to local circum- 
stances. There are widely reported problems with this 
type of structure.29 Such organizations are very likely 
to become an arena for a power struggle between 
local interests and national policies, often between 
local conservative and national reformist forces. In 
these situations, since a degree of collaboration 
between the interest groups is required as a condition 
for the continuation of the program, it is frequently 
the case that one group coopts another, although who 
is coopting whom is sometimes difficult to tell and 
may vary from one locale to another. In the case of 
PSROs, sometimes local conservative physician 
groups might "take over" the PSRO, blunt the reform 
interests of management and carry out the required 
changes with minimum and very deliberate speed. In 
other cases, reformist-minded physicians, armed and 
funded by Federal mandates, might use the PSRO 
as an instrument to advance medical reforms and 
attempt to close down programs which appear to be 
of marginal or dubious quality. This type of struggle 
for control of the organization appears to be an 
inevitable feature of decentralized Federal systems 
of this type. 

7.4.2.2 EFFECTIVE INDICATORS 

When using systems criteria to evaluate the 
effectiveness of PSRO organizations, one arrives at 
a very different set of measures of effectiveness. 
Such indicators, applied at the local level, might 
include: 

• stability or increases of funding (in constant 
dollars) 

• success in competing for demonstration 
projects or other competitive contracts with the 
government 

• success in obtaining contracts with private 
groups to provide monitoring or quality assurance 
services 

• absence of crises in local operations 

• support of local medical societies 



E. A. Krause, "Functions of a Bureaucratic Ideology: 
Citizens Participation," Social Problems 16 (Fall 1968), 
pp. 129-143; 

P. Marris and IVi. Rein, Dilemmas of Social Reform. New 
York: Atherton Press, 1969; 

D. P. Moynihan, Maximum Feasible Misunderstanding. 
New York: Free Press, 1970; 

J. L. Sundquist, Making Federalism Work. Washington 
D.C.: Brookings Institution, 1969. 



150 



• high approval ratings from regional project 
officers. 

High performance on these criteria might be expected 
to be associated with survival as an organization 
in a complicated and often hostile political context. 

7.4.3 Setting Realistic Performance Standards 

Attempts to assess the effectiveness of PSROs 
need to take into account several considerations. 
First, the organizations are quite new: new both in 
the sense that the entire system has only recently 
been created by Federal legislation and many of the 
specified units are not yet or are just in the process 
of becoming operational, and new in the sense that 
the system represents a new type of organization. 
The process of creating new roles and relations, 
new training mechanisms and recruitment and career 
paths does not occur rapidly or easily. 

Second, organizations such as PSROs confront a 
sufficiently difficult and varied set of goals and 
conditions that it is inappropriate and inadvisable 
to settle on any one or even a few criteria in 
evaluating their performance. The assessment of cost 
and quality aspects of health care systems is 
sufficiently complex and poorly understood that it is 
inappropriate to attempt to impose on all PSROs 
uniform procedures and requirements. Further, the 
vulnerability and marginality of these new organizations 
is such that it is inadvisable to specify one or a few 
very concrete indicators of performance — for example, 
reducing days of care in hospitals — because weak 
organizations facing complex goals will all too 
willingly embrace simplistic indicators of performance. 

Finally, it would seem appropriate to consider the 
option of allowing effectiveness criteria to vary 
over time. One might arbitrarily identify three phases 
of PSRO development, each of which might be 
associated with a differing set of effectiveness 
criteria. Phase I refers to the initial start-up and 
"institution-founding" period. Effectiveness criteria 
at this stage would be primarily process measures 
of systems goals: establishing legitimacy of the new 
organization; acquiring sufficient support from local 
constituencies to begin operations; gaining access to 
and entering into agreements with local service 



units; meeting the structural specifications of the 
regional office; getting the data gathering and 
processing units into place, etc. Phase II occurs 
after the PSRO has been in place long enough to 
have established an adequate data processing 
system and relatively secure connections with the 
operational and support organizations in its environ- 
ment. Again, process criteria dominate, but they 
include more measures of goal effectiveness: adequacy 
of data gathering components, the quality of training 
programs for review coordinators and physician 
advisors, technical assistance and educational services 
provided to hospital UR committees, etc. Given the 
need to acquire and retain the support of local 
physicians and medical organizations, educational 
and technical assistance may be more effective 
in the long run than sanctions and denials of 
reimbursement decisions. Outcome indicators appro- 
priate at this stage of development might include 
improvements in the performance of hospitals that 
exhibit extreme deviation from the norm of care or 
cost standards as revealed by profile analyses. Phase 
III criteria are those which are appropriate for the 
evaluation of a "mature" PSRO. In these systems, 
one would expect to observe more sophisticated data 
gathering and analysis systems — e.g., statistically 
supported justifications for focusing decisions — and 
arrangements for conducting and interpreting area- 
wide MCEs and profile studies. Case-adjusted 
measures of cost and quality outcomes should be 
available for selected diagnostic categories. One 
would expect to see some progress in moving toward 
national care standards; but one would also expect 
to find continuing variance around these norms 
reflecting varying local conditions. Local PSROs 
would not be forced to choose between effectiveness 
and survival, but would be evaluated in terms of their 
progress toward improving performance while not 
undermining their essential systems of support. 

The development of a technical assistance and 
educational orientation as opposed to a regulatory 
and punitive approach and the creation of positive 
incentives for physician and hospital compliance to 
supplement or replace the negative sanctions now 
in use would go a long way toward assisting PSROs 
to simultaneously pursue their goals and retain their 
support in their local area of operation. 



151 



8. 1979 Program Status and Directions 
8.1 Introduction 

This section presents information and data on 
PSRO implementation and experience during FY 1979 
and subsequent plans for FY 1980. It covers various 
aspects of the program such as program funding, 
data implementation, project officer performance 
ratings, general management initiatives, hospital 
review implementation, and other forms of review. 
It also attempts to address the following isues which 
are raised in conjunction with the budget cut of $7.5 
million from the proposed budget of $152 million for 
FY 1980: 

• the impact and management objectives which 
the program set for FY 1979, the activities 
undertaken to achieve those objectives, and 
the impacts attained; 

• the measurable objectives set for FY 1980 
and the activities planned to achieve those 
objectives; 

• the cost of PSRO review activities, demonstration 
projects, and special initiatives in FY 1979, and 
the projected costs of those activities for FY 
1980; 

• the focused review activities undertaken by 
PSROs in the modification of concurrent review 
in pursuit of the target unit review cost of $8.70. 

Information on PSRO program status was obtained 
through discussions with HSQB personnel and review 
of HSQB documents such as internal memoranda, 
PSRO transmittals, and PSRO grant applications 
and contract proposals.^ The information presented 



^ Grant applications and contract proposals were submitted 
by 192 PSROs, in phases, over a one-year period which 
primarily covered CY 1978. 



is intended to (1) provide a background for the 1979 
program evaluation, (2) facilitate understanding of 
PSRO activities, and (3) indicate program directions 
for FY 1980 at both the regional and national levels. 
Selected descriptive variables pertaining to the 
implementation of individual PSROs are presented 
in Appendix I. 



8.2 Implementation Status 

8.2.1 Program Funding 

PSRO administrative budgets are funded by annual 
HEW appropriations in the form of grants; hospital 
review activities, however, are funded from the 
Medicare Trust Funds according to Section 112 of 
P.L. 94-182. Table 82 shows the level of PSRO funding 
over the years. 

In implementing the law, the costs for PSRO 
operation are divided into four parts. HCFA 
Appropriated Funds cover Part I of the budget which 
include costs for program management, program 
support, long-term care review, ambulatory care 
review, and special initiatives. The costs associated 
with non-delegated hospital review (Part 11), areawide 
support of hospital review (Part ill), and delegated 
hospital review (Part IV) are reimbursed from Medicare 
Trust Funds. The law also makes provisions for the 
transfer of funds for Medicaid's share of the estimated 
cost of hospital review. Table 83 presents the FY 
1979 funding situation for the ten PSRO regions by 
budget parts; the unit costs shown are for hospital 
review only (Parts II through IV). 

The FY 1980 budget is the result of a cut of $7.5 
million from the proposed budget of $152 million. To 
assure that the operation of the PSRO program will 



TABLE 82 
PSRO Program Funding* 



Periods 



Total 

Program 

Level 



Trust 
Funds 



Appropria- 
tions 



FY 1973 


$ 4,475,000 


— 


$ 4,475,000 


FY 1974 


32,850,000 


— 


32,850,000 


FY 1975 


36,208,000 


— 


36,208,000 


FY 1976 


47,645,000 


— 


47,645,000 


Transitional 








Quarter 


11,977,000 


— 


11,977,000 


FY 1977 


103,000,000 


$41,000,000 


62,000,000 


FY 1978 


147,234,000 


75,000,000 


72,234,000 


FY 1979 


149,866,000 


85,000,000 


64,866,000 


FY 1980 


144,416,000 


85,766,000 


58,650,000 



DHEW, PSRO Fact Book, 1978, plus update by HSQB personnel. 



152 



« 

M 

C 
O 

'bb 
« 

O 

cc 

M 

Q. 

c 
a> 

I- 
« 

00 ^ 
UJ •S 

M 
(A 

< 

bfi 

c 









--SO 



I- (0 08 
Q 



— ^ CO — 



03 
CL 



1^ 




C 

ra o 



« 

o -^ 

CO 

WE 



2. 3 « > 

E °= 
< 



o ^ 



CO • o 

Eo)ES 

£2w 



c 
o 

GC 



<N 





in 


CO 


CM 


in 


■^ 


h- 


1^ 


in 


00 




06 


CT> 


06 


00 


00 


06 


«> 












in 


00 


in 


Oi 


CO 


CD 


00 




in 


CO 


■>r 




<N 


5 





'a-^ 


00 


m 


cd" 


t--' 


00" 


T-" 


co" 


Oi 


C\J 


■^ 







^ 


CO 


CO 


co_ 


cvj 


CVJ 


•^ 


CO 




T- 


^ 


cm" 


cm" 




i/> 












tT 


N- 






h- 


■* 


CO 


t^ 


5 


s 




TT 


CO 


CO 


in 


00 


?^ 


in 














r-" 


ctT 


O)" 


t~~" 


00" 


co" 


TT 


h~ 


in 


h~ 


T- 




r^ 


r^ 


CO 


r^ 


co_ 


CM 


cvT 


cvT 


0" 


«o" 


T-^ 


in" 




-r- 






CM 




<A 















in 


CO 


O) 


CM 





CO 




r^ 


CM 


8 


r^ 


(D 


r^ 


r^ 





r-_ 














cnT 


co" 


0" 


co" 


•t" 


cm" 


<D 


h- 


■^ 


CO 


CD 


CD 


■<3- 


Ci 




co_ 


T— 


00 


■*" 


00' 


00" 


^ 


co" 


■^" 


</J 












1 




C\J 


1 





1 


1 


5 


m 


1 







c6 


CO 


0C5 


CD 








1 






1 




1 




co" 


^ 




0" 




t 


in 


CO 


1 


T— 


1 




CO 


CO 




'" 




</* 












CD 


•"S- 





N- 


CO 


1 


CO 





CM 


co 


■* 


1 


CO 


■■a- 


00_ 


in 


■«t 





co" 


en" 


h-" 


CD 


^ 


1 


O) 


■«t 


CO 


CO 


^ 


' 




in 


C\J 


in 


■t 




<A 














■^ 








1 






CO 







1 


^ 





<J5 





r^ 





CD 




T— " 


0" 


co" 


1 


r-" 




T— 


C\J 


CM 


1 


CO 




■* 


T— 








0» 












CO 





CO 


CO 


O) 


■>* 


CX3 











t- 





t^ 





r-- 


in 


q^ 


h-_ 


co" 


r-" 


00" 


in" 


in" 


CM 


00 


CO 


r^ 


in 


CO 


in 


in 


en 


■*_ 


in 







T-" 




T-" 


co" 


co" 




</J 

















in 


01 


r^ 


in 


5 





en 


r^ 


r^ 


CM 







CO_ 


CM 


■* 


"^ 


in 


in" 


T-" 


cm" 


r^" 


cm" 


CO 


OJ 


00 




in 





CD 




CD 


CM 


in 


00_ 


CJ 


co" 


in" 


r-" 


CD" 


T-" 


«*9 
















_ 


> 


> 


> 



CD 
CO 
00 



O 
CO 
CO 

00" 
in 



CO 

in" 

CO 



T- CO 



O 

8 

in* 

CD 



o 
00 
o 

o" 



i 



i 



t^ 


><«■ 


CO 




CO 


T— 


^ 


t-~" 


00 


in 


CO 


in 



CO 

cd" 

00 



i 



00 



I 



i 



i 



i 



CO 

00 



o 

CD 
CO 
h~" 
CO 
CD 

cd" 



CM 
CM 

o" 

CO 

h~" 

00 



CM 
CM 
CO 

in" 

CO 
CD 



CO 
CD 



in 

CD 



00 
co" 

CD 

■^_ 
cm" 



o 
CO 

CO 

m" 
o 

CD 



CD 

in 
eg 

CO 
CD 

cm" 



CM 

o 

CD 

co" 

00 

o 

■^" 



CO 

c 



•D 




0) 


•0 




C 


TO 


(0 


05 




0) 


5 


0) 


<D 


T3 


■> 


' 


<D 



c 
o _ 

C CO 



_ o 



o 
a> 

Q. 

<n 

0) 



a. 

o 
(0 j: 



o • CO 

S > ^ 

£:^> "75 o 
« C ^ -O Q. <u 



o o 



> 
o 



I 5 
CO CO 









— M* 


J3 




CD 


5 


w to 



> 


Q. 




3 


'> 

CD 

« 

CO 


2 <D 


CO 











E 


-£ ^ 


a. 




*-* 




S 1- 


73 




^ 


<o 


T3 *J 


Q) 




^^ 


tSt 


0) 0) 







CO 


c 


^ 




E 





^ 3 
CO -O 


CD 









in "D 
CO 0) 
g c 


c 


m 




If) 


Q. 
Q. 

3 


.E en 
■0 CO 

1^ 


CD 
CD 
T3 




CO 

X 

■0 

c 
CO 


to 



CO 

•0 


CO 

■a 

c 

CO 

c 


*^ 




CO 


_3 


(U 


CO T3 


C 








E 


C CD 


oc 


C 


(U 


T3 
IS 
< 0) 






CO 


CO 
Q. 

CD 


to 


CO 

c 

CO 

E 



153 



be within the budget limit, the following steps have 
been taken: 

• In FY 1980, PSROs cannot expand hospital 
review implementation until further notice. 

• A concentrated effort will be exerted to provide 
the necessary control to assure that PSROs 
achieve the average unit cost per Federal 
discharge of $8.70. HSQB estimates that PSROs 
will have to focus out approximately two-thirds 
of all reviews to achieve this budget level. 

• PSROs demonstrating poor performance will be 
subject to defunding or budget cuts. 

8.2.2 Physician Membership 

Table 84 reflects the status of PSRO physician 
membership as of the date of the most recent PSRO 
grant application or contract proposal. Data 
concerning this issue were available from 184 of 
the 192 funded PSROs. 

A review of the data shows that: 

• Region V has the largest pool of eligible 
physicians (65,131); Region VIII has the smallest 
(9,252). 

• Region I has the highest percentage of physicians 
who are members (65.8%); Region VI, the 
smallest (26.2%). 

• On average, 49,3 percent of eligible physicians 
are members of PSROs. Three regions (IV, VI, 
and IX) are below the average; seven regions 
are above (see Figure 25). 



Table 85 presents the percentage distribution of 
eligible physicians who are members; 123 PSROs, 
roughly two-thirds of those for whom data were 
available, fall in the range between 40 and 70 percent. 



8.2.3 Review Implementation 

Table 86 shows the PSRO review implementation 
activity by DHEW region as of June 30, 1979. 

Table 87 shows the percentage of total hospitals 
participating in the Medicare program between the 
years 1971 and 1978 and the corresponding physician 
assignment rates based on charges. Concerns have 
been raised that the participation of both hospitals 
in the Medicare program and physicians in receiving 
assignments could have been decreased in response 
to the PSRO program. However, no significant change 
in the rates before and after the PSRO implementation 
in 1975 is shown in the table. 



8.2.4 Area Redesignation 

Redesignation of the original 203 PSRO areas 
based on prevailing medical service patterns has 
resulted in 195 existing areas across the nation. 
During the past year, changes in area designations 
have been minor, including the redesignation of 
counties from one PSRO area to another PSRO area 
within the same State or the assignment of previously 
unassiqned zip codes to existing PSRO areas. 
Table 88 shows the number of designated PSRO 
areas by State. 



TABLE 84 
Summary of Funded PSRO Physician Membership 



Region 



Number of 
Funded PSROs 



Number of 

PSROs 
for which 
Data are 
Available 



Number of 

Physicians 

Eligible 



Number 
who are 
Members 



Percent 
who are 
Members 



I 

II 

III 

IV 

V* 

VI** 

VII 

VIII 

IX 

X 



Total 



13 

27 

27 

25 

42 

8 

7 

6 

32 

5 

192 



12 

27 

24 

22 

41 

8 

7 

6 

32 

5 

184 



20,570 
58,929 
36,236 
36,875 
65,131 
31,338 
14,757 
9,252 
55,248 
11,860 

340,196 



13,536 

30,816 

19,570 

17,943 

32,990 

8,219 

7,598 

5,573 

24,600 

7,021 

167,866 



65.8% 

52.3 

54.0 

48.7 

50.6 

26.2 

51.5 

60.2 

44.5 

59.2 

49.3% 



NOTE: These numbers are based on the most recent grant 
applications and contract proposals available for 
each PSRO. Since physicians may be members 
of more than one PSRO, there may be some 
duplication In the membership count. 



One PSRO in Ohio and **three PSROs in Louisiana are 
alternate PSROs which are not required to have at least 
25% physician membership. 



154 



Figure 25 
Physician Membership 



REGION 



II 



III 



IV 



V 



VI 



VII 



VIII 



IX 



X 

National 
Average 



PERCE NT OF ELIGIBLES WHO ARE MEMBERS 
1 1 — 



20 



40 



26.2% 



— r 
60 



r 
80 



65.8% 



52.3% 



54.0% 



48.7% 



50.6% 



51.5% 



60.2% 



44.5% 
I 
I 



59.2% 



49.3% 



100 



SOURCE: Grant Applications and Contract Proposals 



155 



TABLE 85 

PSRO Review Implementation Activity by DHEW Region 
(as of June 1979) 











Number 




Number 


Number 


Percent 


of 




of 


of 


of PSRO 


Hospi- 


DHEW 


Desig- 


PSROs 


Areas 


tals 


Regions 


nated 


Conduct- 


Con- 


in 




PSRO 


ing 


ducting 


Condi- 




Areas 


Review 


Review 


tional 
Areas 



Number 

of 
Hospi- 
tals 
Under 
PSRO 
Review 



Percent 

of 
Hospi- 
tals 
Imple- 
mented 

in 
Condi- 
tional 
Areas 



Percent 


Percent 


Percent 


of 


of 


of 


Hospi- 


Hospi- 


Hospi- 


tals 


tals 


tals 


Fully 


Partially 


Non- 


Dele- 


Dele- 


Dele- 


gated 


gated 


gated 



1 


13 


13 


100 


274 


269 


98 


87 


6 


7 




II 


27 


26 


96 


424 


390 


92 


83 


6 


11 




III 


27 


27 


100 


509 


449 


88 


84 


12 


4 




IV 


27 


24 


89 


1033 


719 


70 


61 


9 


49 




V 


42 


41 


98 


1113 


908 


82 


82 


1 


17 




VI 


8 


7 


88 


436 


331 


76 


76 





24 




VII 


8 


7 


88 


440 


387 


88 


92 


3 


5 




VIII 


6 


6 


100 


357 


354 


99 


18 


10 


73 




IX 


32 


31 


97 


675 


471 


70 


60 


13 


24 




X 


5 


5 


100 


263 


262 


100 


77 


2 


21 




Total 


195 


187 


96 


5524 


4540 


82 


72 


6 


21 





TABLE 86 
Distribution of Physician Membership 







Number 






















Region 


Number 

of 
Funded 
PSROs 


of 
PSROs, 
Data - 
Avail- 




Number of PSROs with Indicated Percentage of Eligible Physicians 
Who are Members 




0-10 


11-20 


21-30 


31-40 


41-50 


51-60 


61-70 


71-80 


81-90 


91-100 






able 






















1 


13 


12 












4 


2 


3 


2 


1 


II 


27 


27 








4 


7 


3 


7 


3 


2 


1 


III 


27 


24 








3 


4 


9 


6 


2 






IV 


25 


22 








5 


4 


7 


5 


1 






V 


42 


41 




1 


1 


5 


12 


12 


4 


6 






VI 


8 


8 




4 


1 


1 




1 


1 








VII 


7 


7 










2 


3 


2 








VIII 


6 


6 




' 






1 


1 


2 


1 




1 


IX 


32 


32 


1 




4 


2 


8 


6 


5 


3 


2 


1 


X 


5 


5 










1 


1 


3 








Total 


192 


184 


1 


5 


6 


20 


39 


47 


37 


19 


6 


4 


Pefcentage 
Within Each 


of PSROs 
1 Group 




0.5 


2.7 


3.3 


10.9 


21.2 


25.5 


20.1 


10.3 


3.3 


2.2 



156 



TABLE 87 

Percentage of Hospitals Participating in l\Aedicare 

Program and Physician Assignment Rates Based 

on Charges 



Year 



1971 
1972 
1973 
1974 
1975 
1976 
1977 
1978 



Percentage of 


Physician 


Hospitals 


Assignment 


Participating in 


Rates 


Medicare 


(Based on 


Program 


Ciiarges) 


88 


53.8 


90 


50.3 


91 


48.1 


91 


47.8 


92 


47.7 


93 


47.6 


92 


48.2 


95 


49.6 



8.2.5 Data Implementation 

By law, the PSRO is responsible for the arrangement 
and maintenance of the regular review of profiles 
of care and services on an ongoing basis. The 
purpose in establishing and maintaining the PSRO 
Hospital Discharge Data Set (PHDDS) is to meet the 
unique needs of the individual PSRO and to allow 
HSQB to monitor PSRO activities. As of October 1979, 
178 PSROs had established HSQB-approved data 
contracts with outside organizations for data 
processing. The PSRO data system status is 
summarized below: 

Number of PSROs at Various Data Implementation 
Stages as of September 1979 




3 Planning 
PSROs 



183 PSROs 
with approved 
Requests for 
Proposal (RFP) 
for data pro- 
cessing 

3 PSROs with- 
out approved 
RFP 



< 



5 PSROs with 
RFP but no 
contract 



178 PSROs 
with current 
contracts 



TABLE 88 



122 PSROs 
submitting at 
least one accept- 
able data tape 

56 PSROs with 
contracts but 
no acceptable 
data tape 



Number of PSRO Areas, by State 



State 



Final 
Areas 



State 



Final 
Areas 



Alabama 




1 


Montana 


Alaska 




1 


Nebraska 


Arizona 




2 


Nevada 


Arkansas 




1 


New Hampshire 


California 




28* 


New Jersey 


Colorado 




1 


New Mexico 


Connecticut 




4 


New York 


Delaware 




1 


North Carolina 


District of Columbia 




1 


North Dakota 


Florida 




12 


Ohio 


Georgia 




1 


Oklahoma 


Hawaii, American Samoa, 


Northern 




Oregon 


Marianas, Guam, and the Trust 




Pennsylvania 


Territories of the Pacific Islands 


1 


Puerto Rico 


Idaho 




1 


Rhode Island 


Illinois 




8 


South Carolina 


Indiana 




7 


South Dakota 


Iowa 




1 


Tennessee 


Kansas 




1 


Texas 


Kentucky 




1 


Utah 


Louisiana 




4 


Vermont 


Maine 




1 


Virgin Islands 


Maryland 




7 


Virginia 


Massachusetts 




5 


Washington 


Michigan 




10 


West Virginia 


Minnesota 




2 


Wisconsin 


Mississippi 




1 


Wyoming 


Missouri 




5 


Total 



1 
1 
1 
1 

8 
1 

17 
8 
1 

13 
1 
2 

12 
1 
1 
1 
1 
2 
1 
1 
1 
1 
5 
1 
1 
2 
1 

195 



* Area redesignation pending. 



157 



During FY 1979, HSQB provided guidelines for 
State survey agencies and PSROs to excinange 
information within statutory confidentiality constraints. 
State survey agencies periodically examine hospitals 
and long-term care facilities to determine if the 
institutions comply with Federal health and safety 
standards for Medicare and Medicaid purposes. 
Identified problem areas are noted in Statements 
of Deficiencies and may be waived by Medicaid 
Agencies based on results of their Medicaid Review and 
Independent Professional Review. This information 
and other data essentia! for assuring quality of care 
will be made available to PSROs. On the other hand, 
PSROs which routinely collect standardized data and 
frequently visit the institutions will provide State 
survey agencies with norms, criteria, and standards 
as well as information from MCEs and profile analysis. 

Recently the U.S. District Court of the District of 
Columbia decided that PSRO information, except 
for that which is patient-specific, is subject to the 
Freedom of Information Act and that none of the 
exemptions of the Act applies to any of the data 
(hospital and physician-specific information and MCE 
studies) requested by a public citizen health 
research group. However, HSQB's policy is that PSRO 
data and information, not including patient identity, 
physician identity, and certain hospital-specific 
information, are provided solely to assist (1) Federal 
and State agencies in identifying and investigating 
cases or patterns of fraud and abuse, and /or (2) the 
Secretary and Federal and State agencies in health 
planning and related activities. This policy led to 
the decision by DHEW to appeal the District Court 
of the District of Columbia decision. 

8.2.6 National Professional Standards Review 
Council (NPSRC) 

The PSRO legislation mandated the establishment 
of a National Professional Standards Review Council 
composed of eleven physicians of recognized 
standing and distinction in the appraisal of medical 
practice. The Council is to advise the Secretary on 
policy matters pertaining to the PSRO program, to 
provide for the development and distribution of 
information to PSROs and Statewide Councils, to 
review the operations of PSROs and Statewide 
Councils, and to review regional norms of care used 
by PSROs. 

Prospective members are recommended by national 
organizations representing practicing physicians, 
consumer groups, and other health care interests; 
members are selected by the Secretary. In FY 1979, 
five new members were appointed to the Council. 

During FY 1979, the Council dealt with numerous 
issues pertaining to the appropriateness and quality 
of care. Panel discussion on topics such as long-term 
care, PSRO delegation versus non-delegation. National 
Health Insurance, cost containment, health planning, 
PSRO data disclosure and confidentiality were held 
for the exchange of expert opinions and information. 



The Council reviewed diagnostic-specific criteria for 
admission and expected lengths-of-stay for surgical 
procedures, short stay hospital services, and specialty 
hospital services. In addition, the Council requested 
a benefit-cost analysis of diagnostic-specific versus 
system-specific methodologies for concurrent review. 

8.2.7 Statewide Professional Standards Review 
Councils 

According to the PSRO law, Statewide Professional 
Standards Review Councils (SPSRC) are to be 
appointed and established when three or more 
PSROs in the State become conditional. Primary 
duties of the Statewide Councils are to hear appeals 
of PSRO decisions, to review PSRO sanction 
recommendations to the Secretary, to coordinate PSRO 
activities within the State, to disseminate program 
information and data, and, at the Secretary's request, 
to assist in evaluating each PSRO's performance and 
in finding a qualified replacement PSRO, if necessary. 

In compliance with the above, 16 States are 
required to have SPSRCs. As of 1978, a total of eight 
SPSRCs had been contracted in California, 
Connecticut, Maryland, Massachusetts, Missouri, 
New Jersey, New York, and Pennsylvania. By the 
end of 1979, SPSRCs in Michigan and Virginia were 
contracted, and those in North Carolina, Ohio, Florida, 
and Illinois were appointed. An SPSRC is pending 
appointment in Illinois. In Indiana and Louisiana, the 
appointment process has begun. Hence, a total of 
15 SPSRCs are expected to be funded by the end 
of FY 1980. Statewide Council Advisory Groups 
have been established in all Statewide Councils that 
have contracts. Each Advisory Group consists of seven 
to 1 1 members representing hospitals, other health 
care facilities, and health care practitioners other 
than physicians. Besides giving advice and assistance 
to PSROs and Statewide Councils, the members 
work to involve non-physician health teams in PSRO 
activities and provide liaison with other organizations 
and agencies in the delivery and review of health 
care services. 

8.3 Program Components and Management 
Initiatives 

8.3.1 Background 

HSQB has established a set of initiatives to expand 
and improve upon program performance. The 
initiatives described in this section cover three broad 
areas: implementation initiatives and termination of 
funding (8.3.2 and 8.3.3), general management 
initiatives (8.3.4), and review expansion and improve- 
ments (8.3.5 and 8.3.6). 

8.3.2 implementation Initiatives 

The initiative concerning program implementation 
refers to organizational funding and beginning of ' 



158 



review in hospitals. As of October 1979, tinere were 
three planning PSROs, 186 conditional PSROs, and 
six unfunded PSRO areas. The three planning PSROs 
are expected to become conditional sometime 
during FY 1980. Of the 186 conditional PSROs, 
182 are priority conditional and four are alternate 
conditional.- Table 89 shows the number of PSROs, 
by type, at different stages since 1974. 

Conditional PSROs are appraised by the Secretary 
(a responsibility delegated to HSQB) with respect 
to their review effectiveness before they are fully 
designated. The Act, as amended, provides that this 
trial period may not exceed 48 months. However, 
this period may be extended for up to two additional 
years for unusual reasons beyond PSRO control. In 
June, 1979, 10 PSROs reached five years of 
conditional designation, and 52 more PSROs 
reached four years of conditional designation. Hence 
HSQB needs to review the performance of these 
PSROs to determine whether the operation of these 
organizations can be qualified for full designation. 

8.3.3 Termination of Funding 

A major PSRO program initiative in Fiscal Year 
1979 was the discontinuation of financial support for 
ineffective or poorly managed PSROs. Elimination 
of poorly performing PSROs is expected to increase 
the overall cost-effectiveness of the program. During 
FY 1979, the funding of three conditional PSROs 
and one planning PSRO was discontinued. The 
conditional PSROs defunded were the Tennessee 
Foundation for Medical Care, the Calumet Area 
Professional Review Organization, and the Erie Region 
PSRO: the planning PSRO not refunded was the 
Area XIX PSRO in Long Beach, California. In 
addition, the Southern Maryland PSRO chose to 
withdraw from the program. 



■A priority conditional PSRO is an appointed organization 
with at least 25 percent of the practicing physicians in the 
area as members. In areas where the requirement cannot 
be fulfilled by a physicians' organization, other health 
agencies are set up as alternate conditional PSROs. 



When a Regional Office believes that a PSRO is 
experiencing serious performance problems, it notifies 
the PSRO in writing that its grant or contract may 
not be renewed unless substantial improvements 
are made. This warning letter should also make 
technical assistance available to the PSRO and set 
a specific time period in which corrections should be 
implemented. If no improvement results from this 
warning, the Regional Office informs the PSRO of its 
intention not to renew the PSRO's funding and the 
appeals procedures available to conditional PSROs, 
as provided in regulations (42 CFR 463.11). These 
regulations give PSROs an opportunity within 30 
days to request an informal meeting with HEW officials, 
submit written materials, or both, in response to 
the proposed action. If an informal meeting is held, 
the Health Standards and Quality Bureau Director 
designates an official to preside at the meeting and to 
recommend a decision which the Director may 
adopt, revise, or set aside in reaching a final decision. 

8.3.4 General Management Initiatives 

General management initiatives consist of: 

• establishment of performance objectives 
(objective setting) 

• improved financial management 

• improved relationships with external agencies 

• self-assessment of impact 

8.3.4.1 OBJECTIVE SETTING 

Objective setting is regarded by HSQB as the 
cornerstone for HSQB, Regional Project Officers, and 
individual PSROs to improve and document PSRO 
performance. It involves: 

• identification of problem areas, 

• prioritizing activities within funding restraints, 

• setting measurable, specific objectives to resolve 
the problems identified, 

• definition of methodologies to attain the 
objectives, 



TABLE 89 
Implementation Status" 



Type of 
Organization 



7/74 



7/75 



7/76 



7/77 



10/78 



10/79 



Planning PSROs 


91 


58 


33 


62 


15 


3 


Conditional PSROs 


11 


63 


87 


120 


178 


186 


Total 


102 


121 


120 


182 


193 


189 


Unfunded areas 


101 


82 


83 


21 


2 


6 


Total areas 


202 


203 


203 


203 


195 


195 



DHEW, PSRO Fact Book, 1978 plus update by HSQB 
personnel. 



159 



• initiation of corrective actions to solve the 
problems, and 

• monitoring and documentation of progress. 

Therefore, objective setting requires frequent 
evaluation of progress, assessment of barriers and/or 
factors contributing to change, and, when necessary, 
modification of objectives. 

HSQB provides program guidance to PSROs by 
means of transmittals, general memoranda, technical 
assistance documents, workshops, and individual 
on-site visits. Further, explicit criteria for judging 
the acceptability of performance objectives are 
specified as follows: 

• the objective should be measurable, 

• the objective should address the national PSRO 
program goals of improving quality of care and 
fostering appropriate utilization of services, 

• the objective should be related to available data 
that identify major problems in the area, 

• the objective should be realistic in terms of 
effort, required methodology, timeframe, data 
availability, and degree of sophistication. 

• the objective should be related to the results 
and effects of PSRO performance rather than 
the internal operations of the PSRO. 

A PSRO is responsible for analyzing national 
and local data to identify local problems and set 
objectives to improve health care utilization and 
quality in its area. The PSRO should then establish 
specific measurements, methods, and timeframes 
appropriate to its area to achieve the goals and 
objectives which it has set. During FY 1980 a 



subcommittee of the NPSRC was established to work 
with HSQB in developing national goals. 

Grant applications and contract proposals 
submitted by PSROs are required to contain 
proposed statements of impact objectives for 
the coming year. HSQB has recently instituted 
an informal management process wherein all 
proposed objectives are reviewed and evaluated 
against a set of objective criteria. The final 
statements of the processed objectives then become 
the working goals of each PSRO. This process will 
become formalized during the coming year. 
Currently, of the 165 PSROs refunded in FY 1979,^ 
106 PSROs in eight regions have submitted impact 
objectives on an informal basis to HSQB. Three 
planning PSROs and 10 newly conditional PSROs 
are not required to enter this process. Although these 
objectives may not be representative of all PSROs, 
they do provide insights into PSRO goal structures. 
Tables 90 through 94 present the data obtained from 
the Individual Proposed Impact Objective Approval 
Worksheets (HSQB); four components of the work- 
sheets are reflected in these tabulations: 

• Source of baseline data 

• Source of study data 

• Focus of the objective 

• Relationship to proposed national objectives 
A review of data revealed that: 

• A total of 106 PSROs submitted 734 statements. 
Regions III and IV had the largest number of 
PSROs submitting impact objective statements. 



'The rest had funds carried over. 



TABLE 90 
Summary of PSRO Submission Impact Objectives 



Region 











Average 




No. of 


Number of 




Number of 


No. of 


PSROs 


PSROs 


Number of 


Objectives 


PSROs 


Funded 


Submitting 


Objectives 


Per PRSO 




in FY '79 


Objectives 




Plans 
Reviewed 


13 


7 


8 


71 


8.9 


27 


21 


14 


103 


7.4 


27 


22 


21 


135 


6.4 


25 


22 


19 


136 


7.2 


42 


41 


15 


59 


3.9 


8 


6 


7 


45 


6.4 


7 


7 


2 


16 


8.0 


6 


6 


5 


31 


6.2 


32 


29 


11 


121 


11.0 


5 


4 


4 


17 


4.2 



I 

II 
III 

IV 

V 

VI 

VII 

VIII 

IX 

X 



Total 



192 



165 



106 



734 



6.9 



Source: Individual Proposed Impact Objective Approval 
Worksheet (HSQB) revised by HSQB. 



160 



a\ 



m 

< 



M 

a> 

_> 
'J 

|.i 

2 >» 

— n 

O 13 
q: Q 

n 0) 

C 

•D = 

a> 0) 

M l/> 

o re 

O.CQ 

a. o 



re o 

E 

3 
V) 





ro 


0) 


^_i_ 




^ 


« 


o 


*i 


■D 


EQ 


3 




O 


3 


k^. 


o 


2 


*♦" 


Z 


o 


C/3 
(0 






]5 


<D 


0) 

o 




« 


3 


Q. 


3 

o 


O 


Q 



o o £ to t: 

— O CO 



(1) 

o 



■q._ i2 
w (c — 
o *- CD 

i.i 

c 0^ > 
o ■- Q^ 
O cc 



o >- 
O CO O 



E oj o J, £ 

O 3 C ,>< QJ 



LU 

o 






CO 
Q 
Q 

X 
Q. 

< 
Q. 
D 

LU 



^lii 



Q o (D 

w ro S -= 
o *- Q. -^ 



<" 5 o QJ 



c 
o 

O) 


CC 



CVJOOOJCDTtOlD'^OCT) 

■>;)-cocD0O(r3T- •!-■«}■ 



cncocoi-Tri-oocDi- 



00<DC\J<DCM00C0-«TCOCO 
•I- CO CM T- 1- ■* 



CMO)<j)ooir)i-'S-cMCD 

CM T- 



T-OOOOOOOOO 



t-OCJt-OOOOOO 



oooooooooo 



oooooooooo 



00>05COt-t-OC\JCNJO 



'TCDCOOOr^CMCOCMCDT- 

OvJ 1-1- 



OOfCDCOOOOT-OlO 



COOOt^CMCDi-OC\iO)T- 
CO T- ,- 



OO-1-OOOOOOO 



COCMi-OOi-OCO-1- 



00i-CD00t-0i-0 



_ = = >>>>>5<x 

161 



o 

CM 



(J) 
r-' 



CM 



00 

CO 



CO 
CM 





CO 


O 

IS 

C3.C0 

EI 




in 


>. 




CO 


■a ^ 
■n ^ 

(0 ^ 

o <u 

CL.<2 

o > 
a. £ 




(O 


(0 on 




CM 


11 




(/) 


0) 











O 


CO 




il ^ 




3 


3 ir 


00 


O 


CO 5 


T3 


tn 



""5 « 

oo 

•' i" 
S 3 

12 

I? -2 
e o 

11 

E 

3 



^ <M (0 

3 t O 



*i <D p a 
= Q.3Q 

^ O *r 



O O S «J £. 
^ ^ CO 



O 



Q. m 

w (0 ■= 



O C .S 

c ffi 3; 

o »- a> 

O DC 



« c .£5 

C il c *- 



o 



LU 

o 



5^0 



CO 

o 
a 

a. 

cc 
< 
a. 

Q 

LU 



Q. ■- Q. H- 



W (0 ti — 

o *- Q- ;;: 



o 05 

^ — 
Q. 



goo.- 



g 


CC 




162 



TABLE 93 

Summary of Proposed PSRO Impact Objectives 

Focus of Objectives 

by Region 



Region 


Utilization 


Implementa- 
tion 


Cost 


Ancillary 


Quality 


Process/ 
Internal 


Total 












Operation 




1 


28 


3 


8 


1 


1 


30 


17 


II 


34 


8 


8 


4 


7 


42 


103 


III 


64 


4 


15 





8 


44 


135 


IV 


75 


5 


8 


2 


10 


36 


136 


V 


36 





6 





3 


14 


59 


VI 


13 


5 


2 


1 


2 


22 


45 


VII 


6 





2 








8 


16 


VIII 


20 





4 


2 


2 


3 


31 


IX 


89 


1 


1 


2 


5 


22 


121 


X 


7 





4 








6 


17 


Total 


372 


26 


58 


12 


38 


227 


734 


Pet. of All 
















Objectives 


50.7 


3.5 


7.9 


1.6 


~5.2 


31.0 


100.0 



Source: Individual Proposed Impact Objective Approval 
Worksheet (HSQB) Revised by HSQB. 



• With respect to source of baseline data 
(Table 91): 

. . 81 statements (25.6%) indicated that 

MEDPAR ' would be utilized. 
. . 86 statements (27.1 %) indicated that "other" 

data, including PHDDS, would be utilized. 
. . 56 statements specified multiple data source. 

• With respect to source of study data (Table 92): 
. . 170 statements (32.9%) indicated study data 

should be provided through concurrent reviews. 

86 statements (6.6%) indicated study data 
would be provided through MCE studies. 
. . 77 statements (14.9%) indicated "other" 
sources, including focused concurrent 
review, profile analysis, and preadmission 
certification would be utilized. 

• With respect to focus of objective (Table 93): 
. . 372 statements (50.7%) indicated that 

utilization would be their target. 
. . 227 statements (31.0%) were targeted at 
process or internal operations. 

• With respect to relationship to national objectives 
(Table 94): 

. . 480 statements (65.4%) stated that there was 
such a relationship. 



TABLE 94 

Summary of Proposed PSRO Impact Objectives 

Related to National Objectives 

by Region 



' MEDPARzrMedicare Provider Analysis and Review, a 
report series based on the Medicare 20 Percent Discharge 
File (see Section 2.3). 





Number 




Percentage 




of 




of 




PSRO 


Total 

Niimhpr 


PSRO 




Objectives 


Objectives 


Region 


Relating 


of 
Objectives 


Set in 




to 
Proposed 


Relation 
to 




National 




National 




Goals 




Goals 


1 


38 


71 


53.5 


II 


53 


103 


51.4 


III 


87 


135 


64.4 


IV 


95 


136 


69.8 


V 


45 


59 


76.3 


VI 


18 


45 


40.0 


VII 


8 


16 


50.0 


VIM 


28 


31 


90.3 


IX 


97 


121 


80.2 


X 


11 


17 


64.7 


Total 


480 


734 


65.4 



Source: Individual Proposed Impact Objective Approval 
Worksheet (HSQB). Revised by HSQB. 



163 



8.3.4.2 FINANCIAL MANAGEMENT 



8.3.4.2.1 Unit Review Costs 



Effective financial management by PSROs is felt 
by HSQB to be essential for controlling cost and 
assuring that PSRO review is conducted in an effective 
and efficient manner. While this responsibility rests 
with the PSROs, it is incumbent on the HSQB Central 
and Regional Offices to provide the tools and technical 
assistance necessary for accomplishment of this 
task. 

Since 1978, the funding mechanism of PSROs 
has been changed from contract to grant. This 
affords the PSROs a greater degree of flexibility 
to accomplish local objectives within the confines of 
the grant regulations, program guidelines, and the 
total grant award. With this flexibility, the Project 
Officers must monitor their projects closely in order 
to assure that decisions made by the PSRO are 
within the regulations and guidelines on allowable 
cost. 

To assure the financial integrity of PSROs, they 
are required to have an annual audit by an independent 
Certified Public Accountant. This audit encompasses 
both financial management and technical performance. 
These audits are reviewed by the HEW Audit Agency 
for completeness and accuracy. If necessary, the 
PSRO prepares a corrective action plan which is 
monitored by the Project Officer to assure that the 
plan is carried out. In addition, the HEW Audit 
Agency will conduct either a quick assessment or 
full audit to assure proper use of Federal funds. 
Costs determined by any of these audits to be 
unallowable will be recommended for recovery to the 
Contract or Grant Officer and attempts will be made 
to recoup these funds from the PSRO. 

To supplement the audit services available from 
the HEW Audit Agency, HSQB has a contract to 
perform close-out audits of the contracts as the 
funding mechanism is converted to grants and to 
perform timely audits of PSROs where financial 
mismanagement may be suspected. 

Technical assistance on all aspects of financial 
management is available from HSQB Central and 
Regional Offices. A contract is planned for FY 1980 
to assist the Regional Project Officers in their 
monitoring activities by assuring that their knowledge 
of financial management and accounting is current. 



A primary goal of the PSRO program has been the 
reduction of the unit cost of hospital review (from 
$12.31 per discharge in 1977 to $8.70). Of all PRSOs 
funded in FY 1978, an average unit cost of $12.91 
was calculated. This includes costs for concurrent 
review (CR), MCE studies, and areawide review. Of 
all PSROs refunded in FY 1979, an average unit 
cost of $8.77 was negotiated. Previously, additional 
funding was allowed when the difference between the 
actual and the assigned unit review cost was 
justified. However, due to the budget cut, it is 
estimated that two-thirds of the PSRO review activities 
will have to be exempted in order to meet the 
target unit cost. 

HSQB developed an allocation level for each region 
based on funds available for the PSRO program 
in FY 1979. In determining specific regional allocation 
for PSRO review costs, PSROs with greater problems 
in utilization were allocated more funds. Since the 
potential for cost savings is greater where utilization 
levels are higher and vice versa, the rates recom- 
mended to the regions for individual PSROs were 
adjusted 1 percent above or below the average rate 
of $8.70, based on Medicare utilization information. 
Based on this approch, each region was allocated 
a total dollar amount for review costs and an average 
unit cost as indicated in Table 95. 

In their effort to reduce the costs of hospital review, 
HSQB has exerted greater control over reimbursement 
of review costs to hospitals. A special condition has 
been included in PSRO grants which sets a target 
unit cost for both delegated and non-delegated 
review in each PSRO. The PSRO has the responsibility 
for assuring that the negotiated delegated review 
cost is also within this target. PSROs are encouraged 
to monitor the payments made by the fiscal 
intermediary to assure that the cost paid does not 
exceed the target. Fiscal intermediaries are also 
required to consult with PSROs if final review costs 
requested by delegated hospitals exceed that approved 
by the PSRO. 

8.3.4.2.2 Overall Budget 

HSQB has begun to provide each PSRO with an 
overall budget for the costs of hospital review in its 



TABLE 95 
Average Unit Review Cost Assigned to Each Region 



Region 



IV 



VI 



VII 



VIII 



IX 



National 
Average 



Average 


























Unit 


FY 79 


$8.42 


9.61 


9.11 


8.71 


9.24 


8.53 


9.17 


8.45 


8.51 


8.14 


8.94 


Review 


























Cost 


FY 80 


$8.67 


8.90 


8.78 


8.56 


8.83 


8.65 


9.13 


8.60 


8.06 


8.00 


8.70 



164 



area. Within these cost parameters, individual PSROs 
would develop plans including estimated costs for 
the conduct of delegated and non-delegated review. 
Each delegated hospital would develop a budget based 
on the specific review objectives negotiated with 
the PSRO and the estimated workload necessary to 
meet those objectives. The workload would include 
the estimated number of admissions and the manner 
in which they would be reviewed. It would also 
include costs of delegated activities such as 
concurrent and preadmission review and medical 
care evaluation studies. The negotiated budget, 
resulting from revision of the budget proposal, 
would become a ceiling on the total reimbursement 
to the hospital. Analogous standards are applied to 
non-delegated hospitals which are reviewed by 
local PSROs. 

8.3.4.3 RELATIONSHIPS WITH EXTERNAL AGENCIES 

Coordination with external agencies is essential 
for PSROs to operate and monitor Federal health 
care programs. The external agencies with which the 
PSRO program has attempted to establish cooperative 
relationships include: Medicare fiscal intermediaries, 
State Medicaid Agencies, End-Stage Renal Disease 
(ESRD) Medical Review Boards, State governor's 
offices. Health Systems Agencies( HSA) and private 
health insurance programs. 

Table 96 shows the number of PSROs with at least 
one executed administrative agreement in various 
applicable categories as indicated by a review of the 
current grant application or contract proposal. 



It is possible that PSROs can have more than one 
agreement in each area depending upon the number 
of applicable organizations. This is particularly true 
with Title XVIII fiscal intermediaries, Title XIX and 
Title V State Agencies, and HSAs. In addition, Title 
V, HSA, nontitled review, and ESRD agreements 
are not applicable to all PSROs. Since the data are 
not uniformly reported, this table does not reflect 
either pending agreements, category applicability, or 
the potential number of agreements possible within 
a category. 
A review of the data indicates that: 

• 160 PSROs (83.3%) reported an executed Title 
Title XVIII agreement with Medicare fiscal 
intermediaries; 

• 139 PSROs (79.4%) reported Title XIX agreements 
with State Agencies; 

• 47 PSROs (24.5%) reported Title V agreements 
with State Agencies; 

• 84 PSROs (43.8%) reported Health Systems 
Agency agreements; and 

• 25 (13.0%) and 13 (6.8%) reported similar 
agreements for nontitled review and End Stage 
Renal Disease, respectively. 

8.3.4.3.1 Relationships with Medicare Fiscal 
Intermediaries and Carriers 

The Title XVIII (Medicare) program makes financial 
reimbursement through contracted fiscal intermediaries 
(FIs) and carriers in various geographic areas; as 
of March 1979, 296 Memoranda of Understanding 



TABLE 96 
Summary of Reported Administrative Agreements by Region 





Number 
of 






Category of Agreement 






Region 


Funded 
PSROs 
(1979) 


Title 
XVIII 


Title 
XIX 


Title 
V 


HSA 


Non- 
Titled 
Review 


ESRD 


1 


13 


11 


10 


2 


5 


2 


1 


II 


27 


24 


22 


2 


11 


2 


1 


III 


27 


24 


23 


12 


11 


4 


4 


IV 


25 


21 


20 


8 


14 


3 


1 


V 


42 


33 


31 


7 


16 


2 


1 


VI 


8 


7 


7 


5 


3 


1 





VII 


7 


5 


5 


2 


5 


1 





VIM 


6 


6 


6 


5 


5 


4 





IX 


32 


25 


11 


1 


10 


6 


5 


X 


5 


4 


4 


3 


4 










192 


160 


139 


47 


84 


25 


13 


Pet. of PSROs with 














at Least One 
















Administrative 














Agreement 




83.3 


72.4 


24.5 


43.8 


13.0 


6.8 



Source: Grant Applications and Contract Proposals. 



165 



(MOUs) were written. An intermediary or carrier is 
permitted to issue reimbursements only if tlie PSRO's 
review indicates that the health care services were 
necessary and appropriate. Since an Fl is responsible 
for the final settlement of all delegated hospital 
PSRO review costs, the following measures must be 
taken in order to assure that each PSRO will receive 
full support in operating within its target unit cost 
limit and budgeted limits: 

1. Analysis of actual PSRO review costs when 
cost report is submitted. 

2. Consultation with PSRO to determine the final 
adjustments to the payment. 

3. Payment of interim reimbursement. Advance 
notification to hospitals if its review cost exceeds 
its approved unit cost rate. 

To explore ways of integrating the activities of 
PSROs and Medicare carriers with regard to 
Physician Service Insurance (Part B), plans for 
demonstration projects were designed for the PSROs 
in Washington State, Montana, California XXIII, 
Western Massachusetts, and Highlands, Pennsylvania. 

FIs are also required initially to monitor PSRO 
review decisions on a 20 percent sample of all PSRO 
reviews. HSQB recomended that an Fl monitor PSRO 
performance on a hospital-specific basis for at least 
six months after the PSRO has begun review in that 
institution. If, at the end of the six month period, 
the rate of disagreement between the Fl and PSRO 
is less than 3 percent of the cases reviewed, Fl 
monitoring will be discontinued. Otherwise, a decision 
concerning the continuation and form of monitoring 
would be made by the HSQB Regional Officer, in 
consultation with the Medicare Regional Director. 

Available information indicates that FIs show 
varying degrees of disagreement with PSRO decisions. 
The disagreement ranges from less than 5 percent 
of the sample cases to almost 25 percent, with most 
PSROs at the lower end of the range. Where 
disagreements exist, the differences are resolved 
between the two organizations. 

8.3.4.3.2 Relationships with State Medicaid Agencies 

The Act provides for State Medicaid Agency 
review of PSRO activities at discrete points during 
the development and implementation of the PSRO 
review. 

Prior to implementation of the PSRO program, each 
Title XIX (Medicaid) State Agency was responsible 
for performing its own quality assurance and 
utilization review on health care services provided to 
Title XIX recipients. When a PSRO is conditionally 
designated, it assumes this responsibility with respect 
to health care services provided in institutions under 
its review. 

According to P.L. 95-142, PSRO review of Medicaid 
services is not conclusive for payment purposes 
unless (1) the PSRO and the State Medicaid Agency 
agree on the administrative arrangement and execute 



a Memorandum of Understanding; or (2) the Secretary 
waives the requirement for an MOD on the basis of 
the State's request or because the State has not 
negotiated in a timely manner or in good faith. As 
of March 1979, 114 MOUs had been signed by 
PSROs and State Medicaid agencies. 

A State may receive Federal Medicaid matching 
funds for the costs of monitoring PSROs under an 
approved plan. It has the opportunity to review and 
comment on a PSRO's annual review plan, monitor 
PSRO review determinations and their effects on 
State expenditures, and appeal to HSQB for 
modifications of PSRO review or removal of a PSRO's 
authority to make funding determinations in particular 
circumstances. HSQB has requested that State 
Medicaid Agencies monitor PSROs by participating 
in objective setting and overseeing Title XIX review. 
At present, HSQB is developing mechanisms to 
promote the cooperation between local PSROs and 
State governments. Table 97 indicates the State 
Medicaid agencies with PSRO-approved monitoring 
plans. 

8.3.4.3.3 Relationships with End-Stage Renal Disease 
Medical Review Board 

IN FY 1978, Congress passed P.L. 95-292 which 
amends the end-stage renal disease (ESRD) 
provisions of the Act in order to achieve more 
effective control of the cost of the program. Thirty-two 
ESRD networks, each with a Medical Review Board 
(MRB), were established throughout the nation to 
coordinate ESRD patient referral to accessible and 
appropriate quality care. 

The ESRD regulations call for the coordination of 
MRB and PSRO responsibilities to avoid duplication 
of activities. Responsibilities of the two systems 
are specified in the statutes and regulations as 
follows: 

• The MRB is responsible for the review of the 
appropriateness and quality of medical care 
provided to ESRD patients throughout the network 
as stipulated in the ESRD regulations. The MRB 

is to develop a review plan to cover all the 
regulatory requirements. 

• The PSRO, by statute, has the responsibility 
for making medical necessity determinations 
for payment purposes in institutional settings 
for which the PSRO has assumed responsibility. 
Therefore, the PSRO has the continuing 
responsibility for performing medical necessity 
review of renal patients who are hospitalized 

or placed in long-term care institutions in its area. 

As of March 1979, 19 MOUs had been signed 
between ESRD MRBs and PSROs. 

In FY 1979, HSQB transmitted guidance to MRBs 
and PSROs so that they can make written arrange- 
ments as to how an individual MRB plans to make 
use of the PSRO's capacities to carry out their 
overall ESRD review plans. 



166 



TABLE 97 

State Medicaid Agencies with Approved 
l\^onitoring Plans from PSROs 



Region 1 


Region VI 


Massachusetts 


New Mexico 


New Hampshire 




Vermont 




Region II 


Region VII 


New York 


Iowa 


New Jersey 


Kansas 




Missouri 


Region III 


Region VIII 


Pennsylvania 


North Dakota 


Maryland 




Region IV 


Region IX 


North Carolina 


California 


Mississippi 


Nevada 


South Carolina 




Region V 


Region X 


Ohio 


None 


Illinois 




Michigan 





8.3.4.3.4 Relationships with State Governor's Officers 

The PSRO relationship with State governor's offices 
was legislatively mandated in the 1977 amendments 
to the PSRO statute. Input from the governor's 
office occurs at five times in a PSRO's development: 

1. Conversion from planning to conditional status, 

2. Conversion from conditional to fully delegated 
status, 

3. Appointment of Statewide Councils, 

4. Assumption of long-term care review, 

5. Assumption of ambulatory care review. 

At each point, the PRSO must simultaneously 
submit its formal plan or modification to the State 
governor and the Secretary of DHEW. The governor 
will submit comments to advise the Secretary. 

8.3.4.3.5 Relationships with Health Systems Agencies 

HSQB and the Bureau of Health Planning, Health 
Resources Administration, Public Health Service, 
have implemented a joint policy statement outlining 
requirements for coordination of PSRO and HSA 
activities in accordance with the provisions of 
P.L. 93-641. The relationship between the local 
organizations is governed by a Memorandum of 
Understanding (MOD) which delineates procedures for 
sharing of community data and joint involvement 



in issues of health planning toward control of health 
care costs. 

In developing its Health Systems Plan (HSP) and 
Annual Implementation Plan (AlP), an HSA has to 
obtain necessary data from a PSRO and coordinate 
its activities with the PSRO to ensure consistent 
policies in their respective activities in the region. 
The HSA can assist the PSRO in maintaining an 
impact on the total health care cost by certifying new 
long-term care beds in cases of need and by 
facilitating the closing of unnecessary acute care beds. 
An HSA can assure quality of care through certifying 
qualified institutions. 

The MOD provides for the sharing of technical 
assistance and other general information. PSRO 
utilization information and MCE studies can identify 
shortages or excesses at various levels of care 
which the HSA can act on. The PSRO can provide 
professional medical judgements to the HSA, 
especially with respect to quality of care, utilization 
of services and facilities, and the need for new 
resources based upon its expertise and experience 
with Medicare/Medicaid patients. Table 98 shows 
the number of MOUs between HSAs and PSROs. 

Currently, although the majority of HSA/PSRO 
MOUs have been signed, there has been a sense 
that the resulting relationships are more "process" 
than "product" oriented. Based on the recently 
collected information, it was found that most PSROs 
and HSAs with MOUs do interact in the minimum 
required ways, i.e., exchange of some data, attendance 
at board and/or committee meetings, and providing 
opportunity for comment on plans, planning criteria, 
and projects. However, only 38 or 16 percent of those 
HSAs and PSROs with MOUs seemed to have 
undertaken the four joint activities that could lead to 
improvement of their local health care delivery system. 
Those activities are categorized as follows: 

• useful data exchange activities; 

• joint development of criteria for need assessment; 

• joint project review; and 

• special issue studies. 

8.3.4.3.6 PSRO Review of Private Insurance Programs 

Employers in private industry paid about $40 
billion of the nation's $183 billion in health care 
expenses in 1978; employee medical benefit costs 
account for as much as 10 percent of total 
compensation in some companies.'^ This situation 
has led many large corporations to collaborate as 
regional groups to develop cost-cutting strategies. 
One of the programs initiated is to increase the use 
of PSROs. An informal survey conducted by the 
American Association of Professional Standards 
Review Organizations (AAPSRO) indicated that as of 
February 1979, 49 PSROs were performing reviews 



Business Week: August 6, 1979, pages 54, 55. 



167 



TABLE 98 
Number of Signed MOUs Between HSAs and PSROs* (FY 1979) 



Organization 
Status 



No. of 
PSROs 



HSAs/SHPDAs* 
which must 
sign l\/IOUs 



No. of 

Potential 

MOUs 



No. of 
Signed 
l\/IOUs 



% Signed 
MOUs 



Planning 

Conditional 

Total 



3 
186 
189 



214 
214 



20 
291 
311 



3 
231 
234 



15% 
79% 
75% 



Information does not include those HSAs and PSROs not 
funded. 



SHPDA=State Health Planning and Development Agency. 
Total reflects the fact that one PSRO is presently 
reviewing two PSRO areas. 



on non-Federal patients. Many PSRO-related 
independent organizations are also performing PSRO- 
type review on private patients. 

8.3.4.4 SELF-ASSESSMENT 

Regional project officers are responsible for 
monitoring PSRO programs at the local level. In FY 
1979, PSROs were required to report areawide 
impact or changes which occurred in the current 
year by submitting impact statements to the Project 
Officers. PSROs are also required to document impact 
related to established objectives as part of their 
annual applications for refunding. 

In view of the need to document PSRO impact, 
AAPSRO summarized impact information from 43 
PSROs. The report covered areas such as utilization, 
quality, and costs of health care for the Federal 
beneficiaries. However, PSROs' methodologies of 
estimating impacts have raised questions as to the 
precision of the impacts claimed. 

8.3.5 Hospital Review Implementation 

The Social Security Act sets forth three major 
PSRO review components in hospitals: 

1. Concurrent review of the medical necessity 
and appropriateness of admission to, and 
continued stay in, a hospital. 

2. Medical care evaluation (MCE) studies to assure 
the quality and improve the nature of the 
utilization of health care services; and 

3. Analysis of health care practitioner, institutional, 
and patient profiles. 

These components of the PSRO hospital review 
system interact to form a comprehensive quality 
assurance mechanism. The compiled data of these 
three components reflect the standards of performance 
of various practice areas. In this way, PSROs can 
focus concurrent review activities on problem areas 
or exempt consistently acceptable areas from review. 



8.3.5.1 CONCURRENT REVIEW 

A PSRO is responsible for assessing, through 
concurrent review, the utilization of services 
involving Federal program patients. The purpose is 
to determine whether: 

1. The services are medically necessary, 

2. The quality of services meet professionally 
recognized standards of health, and 

3. The services are provided in the most appropriate 
setting. 

PSROs achieve these objectives by conducting 
admission and continued stay reviews using PSRO- 
approved health care criteria. Admission reviews 
are required to be completed, and certification or 
adverse notice provided, as soon as feasible, but no 
later than three working days after the admission. 
If the admission is certified as medically necessary, 
an initial length of stay checkpoint is assigned using 
regional or local length of stay norms for a patient 
of the same age with the same diagnosis or problem. 
Through continued stay review a second length of 
stay checkpoint is assigned according to the patient's 
condition, diagnosis and age. Subsequent continued 
stay reviews are performed if the patient is still 
hospitalized at the end of the revised length of 
stay checkpoint. At the point when review of the 
patient demonstrates that further hospital care for 
the patient is no longer necessary or appropriate, 
the attending physician must be notified and given 
an opportunity to discuss the case with the physician 
advisor. If the physician advisor still believes that the 
patient no longer needs hospital care and the 
attending physician disagrees, the PSRO must issue 
a notice of adverse determination to the patient or 
his representative, the attending physician, the 
hospital, and the fiscal agent. This notice serves to 
inform the parties that Medicare or Medicaid 
payments are terminated. If the attending physician 
has ordered a lower level of care for the patient and 
the PSRO agrees, or if in discussion with the 



168 



PSRO the attending physician agrees to discharge 
the patient to extended post-hospital or institutional 
care, the PSRO is responsible for notifying the 
discharge planners in the hospital to assure appropriate 
placement for the patient. 

Where feasible, procedure reviews must also be 
conducted, on admission or during hospitalization, 
before an elective surgery or an elective major 
procedure is performed. PSROs may also perform 
preadmission review to assess the necessity and 
appropriateness of any elective health care service. 

In order to ensure consistent and appropriate 
review standards, a PSRO must provide to each 
hospital in its area a copy of the norms, criteria, 
and standards to be used in preadmission and 
concurrent review, or in an alternative review method. 
A PSRO may use alternative methods for conducting 
review in all or part of the PSRO area if its written 
plan demonstrates potential for equal or more 
efficient and effective review mechanism and is 
approved by HSQB. Several such methods are 
discussed below. 

8.3.5.1.1 Intensity, Severity, and Discharge Screening 
Criteria « 

The Intensity, Severity, and Discharge (ISD) 
alternative to concurrent review varies from the 
traditional review on two basic elements: process 
and criteria. The review is a cyclical process initiated 
within one working day from admission and recurring 
every three days regardless of diagnosis. The 
timing of review is periodic and independent of 
length-of-stay percentiles for diagnoses. 

The ISD criteria are a combination of systems cri- 
teria, level of care criteria, and outcome criteria rather 
than diagnosis-specific criteria. The three intensifica- 
tion components are Severity of Illness, Intensity of 
Services, and Discharge Screening. The Severity 
of Illness component consists of explicit criteria 
applicable to body systems and to the disease process 
as manifested bv objective signs and symptoms 
documented in the medical record. The criteria were 
developed for thirteen body system groupings, such 
as respiratory/chest which correspond to ICD-9 CM 
categories, and a generic group applicable to all 
systems. Pediatric and surgical criteria are not 
included. 

The Intensity of Services component relates to the 
exolicit services ordered and provided which express 
the intensity of medical management. The Discharge 
Screening component of the criteria consists of 
obiective outcomes, such as degree of functional 
independence or progress of the patient toward goals. 



which are expected to be present at discharge and 
indicate readiness for discharge. 

The ISD approach has been used in Alabama, 
Michigan, California, Massachusetts, Florida, and 
South Carolina. Proponents agree this alternative is 
less expensive and less time-consuming than 
traditional review which focuses on diagnostic labels 
and that it allows focusing in on problems and/or 
focusing out of review where acceptable utilization 
patterns exist. Specifically, the PSRO may focus in for 
more intensive review of groups of bodily systems. 

8.3.5.1.2 Centralized Review in Rural Hospitals^ 

A further variation of the ISD procedures and 
criteria was applied in a centralized review system 
in rural and non-delegated hospitals in Alabama 
using a toll-free statewide telephone network. Local 
hospital personnel transmitted medical information 
via the network. Based on the physician progress 
notes, nursing notes, x-ray, and other ancillary 
service results, the central review coordinator applied 
Intensification Criteria to certify the admission, or 
confirm the stay, or identify a case for referral to a 
physician advisor. The impact study report on this 
centralized review documented an increase in 
denials, a decrease in length-of-stay, and a decrease 
in review cost from the traditional review process. 
This experience suggests that ISD criteria can be 
flexibly applied to various types of review processes. 

8.3.5.1.3 Appropriateness Evaluation Protocol (AEP) « 

Developed by Gertman and Restuccia in recognition 
of the need to differentiate reliably between appro- 
priate and inappropriate utilization of medical 
care, the AEP is a protocol intended for utilization 
review. It has been applied both retrospecitvely and 
concurrently in medical care review. 

The AEP criteria are neither diagnostic- nor 
system-specific but incorporate elements of level-of- 
care criteria in the process. The instrument and 
protocol are based on a set of 27 objective 
criteria related to medical services, nursing/life 
support services, and patient condition factors. These 
criteria have been found to be associated with 
appropriate use of a hospital bed. However, the 
protocol allows'the reviewer to incorporate additional 
subjective criteria to override the explicit criteria in 
determining final appropriateness. Although it has 
been argued that the override option increases the 



'The ISD Screening Criteria were developed and copyrighted 
by InterQual using private funds. A Tecfinical Assistance 
Manuai for Application of ISD Screening Criteria was 
adapted for use by PSROs by InterQual under Purchase 
Order No. HCFA-M 78-0640. 



' Impact Study: Topic: Alternative Approach to Concurrent 
Review in Rural Non-Delegated Hospitals. Prepared by 
Barbara Ridgeway, RN of the Alabama Medical Review, Inc. 

'Paul M. Gertman, Appropriateness Evaluation Protocol: 
Development and Methodological Testing of a Nov/ 
Technique for Studying Inappropriate Hospital Utilization, 
Report submitted under Contract No. SSA 6090-75-0209. 



169 



reliability of the AEP instrument, this remains to be 
determined. 

The AEP instrument has had limited methodological 
testing; validity and reliability testing remain to be 
conducted. Several PSROs have used the instrument 
as a means of self-assessment to determine 
independently their review effectiveness. 

8.3.5.2 FOCUSED REVIEW 

Modification of concurrent review through focusing 
is a strategy for PSROs to achieve their individual 
objectives relating to improvements in utilization and 
quality of care, and to improve the efficiency of the 
review system. The mandated reduction of the 
assigned average unit cost for hospital review to 
$8.70 motivated PSROs to expedite the development 
and refinement of focusing methodologies. The 
direction of review modifications is two-fold: (1) 
exemption of review in certain categories of cases 
that are consistently appropriately cared for in 
hospitals, and (2) intensification of review in known 
or suspected problem areas. 

When a PSRO exempts categories of cases from 
admission review and/or continued stay review, 
such decisions could apply to diagnoses, problems, 
or conditions; elective surgical or major diagnostic 
problems; practitioners; and major clinical area(s) 
within a hospital, or total hospital(s). 

It is envisioned that as PSROs exempt categories 
of cases from review, costs per admission for 
concurrent review will decrease. Conversely, as PSROs 
begin to exempt categories, the cases being reviewed 
may be more complex and, therefore, more 
expensive to review. Effectiveness of review is 
expected to be realized by concentrating review 
efforts on identified problem areas. Mechanisms to 
Intensify review include preadmission review, 
preprocedure review, intensification of admission 
and/or continued stay review, and focusing in one 
or more areas related to quality of care and use of 
ancillary services. 

In order for PSROs to decide on the categories for 
focusing, appropriate experience and judgment, an 
adequate computerized data base which allows 
analysis of areawide utilization patterns, and trend 
analysis over time are essential. In addition to 
aggregated data analysis, findings from MCE studies 
of selected retrospective analysis of patient charts 
can assist the PSRO in drawing conclusions about 
appropriate utilization of services. 

Table 99 reflects the status of PSRO focused review 
implementation among PSROs who conduct hospital 
review. The data were obtained from grant application 
and contract proposal review and cover the current 
performance period of some 69 PSROs. The data 
should not be interpreted to mean that other PSROs 
did not perform focused review during their current 
performance period; however, data were not available 
in the funding documents to support the inclusion 
of additional PSROs. 



The extent to which PSROs are performing focused 
review is summarized as follows: 

• 69 (36.9%) of the PSROs conducting hospital 
review clearly had implemented a focused review 
component; 

• 57 PSROs were conducting exemption of review; 
the bulk of this group (52 PSROs or 91.2%) were 
exempting on the basis of diagnoses, problems, 
conditions, or procedures; 

• 43 PSROs were conducting intensification of 
review, with the bulk (32 PSROs or 74.4%) 
intensifying with respect to particular hospitals. 

8.3.5.2.1 Care Level and Timeliness Review (CLTR) » 

CLTR is a periodic retrospective review of a 
statistically representative sample of patient records. 
It is a tool to evaluate the appropriateness of 
admission, the timeliness of care, testing, and services 
provided to a patient, and the timeliness of discharge. 
The CLTR methodology can help PSROs and 
hospitals to determine the extent and underlying cause 
of hospital misutilization, develop corrective actions 
for reducing avoidable hospital days, and evaluate 
progress in implementing these actions. 

CLTR is a supplementary approach to concurrent 
utilization review. The issue of the appropriateness 
or quality of services provided to patients once the 
necessity of admission is determined is not addressed. 
The unique contribution of CLTR is the evaluation 
of the timeliness of care and the efficiency (but not 
the quality) of medical care management. Further, 
CLTR may be useful for focusing in (intensifying) 
concurrent review procedures to concentrate on 
possible problems or focusing out (exempting) 
institutions and providers from concurrent review 
procedures. 

The CLTR has been tested by two PSROs and 
several hospitals with inconsistent findings. The 
methodology has not yet been determined to be 
reliable. Although no correlation was found between 
traditional UR appropriateness determination and 
CLTR appropriateness determination, this supports 
the contention that CLTR, if it is valid, does in fact 
measure a different construct. Economically, the 
CLTR audit was estimated to cost twice as much as 
the average medical care evaluation study conducted 
in PSRO areas where CLTR was tested. 

8.3.5.3 MEDICAL CARE EVALUATION 

The quality of medical care delivered by physicians 
and other health care practitioners is assessed 
by PSRO Medical Care Evaluation Studies (MCEs) 
(see Section 5). MCEs involve peer review of selected 



"McKlnsey and Company, Inc., Technical Assistance 
Document: Implementing a Care Level and Timeliness 
Review Procedure, March 1978; and Testing and Refining 
Care Level and Timeliness Review: Final Report, March 31, 
1978, HEW/HCFA/HSQB, Contract No. HSA 240-76-0071. 



170 



a 

0) 

o 

X 



.J. w 

O 0) 

« S 



ffl « " 2 



9) 



< 



a> 

OQ (V 
0) 

Iff 

3 
U 

o 



c 

Q. 

E 

X 

ai 



Q. 



0)^p 

C 0) 5? 
'— •*-« ^ 

c 



(A 
O 

I 



O 0) 
CO c 



O'co-gS 
.5 <u o 5 

c 



1— (O^lOin-r-COCVJ^T- 



O^t-IOt-Ot-t-COO 



cvir^o^coooi-coo 



COOlOtDCDT-cOCVJCOi- 



OOCNJCslOv-T-T-cOO 



T-T-C\J^OOCOi-COO 



C\JCVIOOCDt-,-c\j^1- 



(O 



3 CO C -1= HI c 5 

2 Q-<3 0CC 



If) 






Q. 

E 
a> 

X 

CD 

c 

(0 

u 



o o O 03 0) ^ 

-9 Q. tr c D) « <!' 



3 -R qI -S •- " 



CO 



o E 



Q) « tj .± 5 
EoQC-Dg'«-j5> 

3 CO c — O H- 0, 
Z Q.O X DC 



C\IC\Jt-00)t-COCM(Ot- 



00OT-;0000CDt— 0(0 
C) O 00 If) (£> 00 N." O CNJ O 

com^^cNcvjinmcMCM 



'a-COCOi-T-CM^COt>^T- 



T- C\J CM CM Tf CO 






o 

cc 

CO 



T3 

c 
o 
tj 

U) 

o 
cr 

CO 
. CL 
m 

(0 "o 
m 
o ■= 

Q. to 

P Q 



III 



- = = ^>>>>><x 



c _ ^ 

CO -5 I 

•^ CO c 

2 E§ 



See 

^ So 

|jj 5 o 
O (D a> 

OC > Q. 

sis 

CO . 



171 



medical topic areas such as medical diagnosis or a 
diagnostic, therapeutic, or surgical procedure. Studies 
usually are performed retrospectively on a sample 
of medical records. 

During FY 1979, HSQB revised the 1977 MCE 
policies and quality review guidelines to coordinate 
them with the quality assurance standards of the 
Joint Commission on the Accreditation of Hospitals 
(JCAH) and to reflect advances in the quality review 
area. The new policy reduces the emphasis on MCE 
process and numbers requirements, stressing instead 
the demonstration of impact through resolution of 
problems in care. Problem resolution will involve 
identification of problem areas, establishment of 
priorities, and analysis of the problem areas. To 
maximize effectiveness, PSROs and hospitals will 
have to coordinate their efforts in addressing local, 
regional, and national problems. 

The previous MCE policy specified 10 steps of 
quality review which were to be followed by PSROs 
and delegated hospitals. Instead, the proposed policy 
encourages innovative study designs to resolve 
identified or potential problems and the use of data 
sources other than the medical record. The elements 
defining the process are: 

• problem identification 

• data gathering and peer analysis 

• intervention 

• follow-up to assure the problem was resolved 
or reduced and 

• documentation and reporting. 

Detailed plans consistent with the proposed policy 
are to be submitted by the PSROs for approval by 
HSQB. 

The Hospital Review Regulations require every 
participating hospital to conduct four to 12 MCE studies 
depending on the number of admissions to the 
hospitals. According to the proposed policy, all but 
small hospitals and specialty hospitals will have to 
conduct four or more studies. PSROs will reimburse 
for four acceptable studies and any additional studies 
agreed upon consistent with PSRO budget limits and 
hosDital problem assessment. 

The proposed PSRO quality review policy is 
summarized in Figure 26. 

8.3.5.4 PROFILE ANALYSIS 

Profile Analysis is a form of retrospective review 
in which aggregated patient care data are subject 
to pattern analysis. A profile is defined as the 
presentation of aggregated data in formats which 
display patterns of health care services over a period 
of time. PSROs are required to perform profile 
analysis with respect to (1) patients, (2) practitioners, 
and (3) institutions. 

As the last component of the hospital review systems 
to be implemented, profiles are performed for the 
following purposes: 

1. to identify patterns of services which may require 
modifications in review activities; 



2. to identify possible topics for MCE studies; 

3. to evaluate PSRO performance, including 
comparison with other areas and regions of 
the country; 

4. to compare the performance of different providers 
(in^itutions or practitioners) in a PSRO area; 

5. to monitor review activities delegated to 
hospitals; 

6. to determine whether the care and services 
ordered or provided were necessary, appropriate, 
and of acceptable quality. 

The process of profile analysis involves: 

1. comparison of patterns of care by similar 
providers (institutions or practitioners); 

2. comparison of current patterns with previous 
patterns; 

3. identification of patterns that deviate from 
established norms, criteria, and standards; 

4. tracking of care provided to particular patients 
or by particular practioners. 

In order to implement these features, PSROs are 
required to be equipped with a sound data capacity. 
As of FY 1979, 123 PSROs reported profile activity. 

Seven PSROs which were advanced in profile 
analysis received special technical assistance during 
FY 1978. The contractor's final report i" indicated 
that progress had been made in the following areas: 

• diagnosis and procedure profiling of focus review; 

• physician profiling of focus review; 

• profiling in support of monitoring; 

• development of profiles for feedback to 
hospitals; 

• profiling in support of both individual hospital 
MCEs and areawide MCEs; 

• profiling to set PSRO objectives; and 

• profiling to define and document problems. 

Various efforts have been undertaken by HSQB 
to assist PSROs in the development of profile 
analysis. HSQB distributed detailed technical 
assistance documents on case studies, focusing, and 
data quality control; feedback on these from 15 
PSROs was encouraged at a HSQB-sponsored 
conference. HSQB set up contracts with approximately 
100 PSROs to develop techniques and methodologies 
in the use of profiles for surgical rate study and in 
the control of case mix. HSQB sponsored four 
regional training sessions and provided 14 PSROs 
with on-site technical assistance. PSROs advanced 
in profiling are encouraged by HSQB to evaluate and 
assist other PSROs in this area. 

8.3.5.5 SURGICAL REVIEW 

The data on surgery show a wide variation in the 
incidence of surgery. In an effort to reduce the 



' InterQual, PSRO Special Activity Projects in Profile 
Analysis; Documentation of Activities, Oct. 1, 1977— 
Sept. 30, 1978. 



172 



Figure 26 
Proposed PSRO Quality Review Policy 



NATIONAL PSRO 
PROORAM 

loerm Fi CATION 

or PROBLEMS 



■<5— •• 



LOCAL. WRO 

IDKNTTFTCATION 

OP PROBLJEMS 



NATIONAL PSRO 
PROGRAM GOALS 



K 



HOSPtTAL 
IOCNT1F1 CATION 
OF PROBLEMS 



PSMO 

oajccnvcs 



PSRO PROBLEM 
LiSr 



i 




PROBLEMS 
RESOLVES 



PROBLEMS 

nor 

RCSOLVeo 



NEW 
PROBLEMS 



RCPOfrr IMPACT 



RCSULT3 



173 



variation in the incidence of surgery, HSQB is involved 
in several projects: 

1. Development and distribution of statistical 
data on surgical utilization to all PSROs. 

2. Dissemination of AMA-developed sample criteria 
sets for reviewing the medical necessity of 
surgical procedures. These criteria will be 
distributed in calendar year 1980. 

3. Initiation and monitoring of a contract for the 
evaluation of PSRO impact on the quality 

and utilization of surgical services in five selected 
PSROs. 

Statistical Data on Surgical Utilization 

Using the Medicare Claims files, HCFA identified 
the 10 most frequently performed surgical procedures 
in the elderly population. Because of the volume of 
surgical rate data that have been prepared, the 
Administrator of HCFA and the Director of HSQB 
decided to distribute the information to PSROs in 
stages. Data on lens extraction, prostatectomy, 
arthroplasty or hip replacement, and cholecystectomy 
have been released. In 1980 data on mastectomy, 
hysterectomy, resection of small intestine or colon, 
inguinal hernia, hemorrhoidectomy, dilation and 
curettage, as well as rates for all procedures 
combined will be released. 1973 data and 1976 data 
have been issued for PSRO areas. States, HEW 
regions, and census regions. PSRO areas with 
surgical rates 20 percent or more above, and those 
with surgical rates 20 percent or more below, the 
average national rate are also identified. HSQB 
expects these data to be useful in focusing surgical 
review activities. 

AM A Surgical Criteria Sets 

HSQB contracted with the American Medical 
Association (AMA) to coordinate a project to develop 
sets of criteria for reviewing the medical necessity and 
appropriateness of surgical procedures. The AMA 
subcontracted with 13 medical and surgical 
specialty organizations for criteria set development 
and for their editorial assistance in the final 
preparation of the criteria for publication and their 
subsequent refinement and revision. Each specialty 
organization addressed only those procedures which 
account for approximately 75 percent of the surgical 
activity within its area of specialization. 

The sample draft surgical criteria sets were 
distributed to all PSROs in 1979 and are intended 
to serve as samples which PSROs can utilize in 
developing and refining criteria for local review 
systems. The sample sets were developed for 
screening large numbers of cases to select cases for 
physician peer review. As screening criteria, they 
do not attempt to provide an exhaustive list of all 
possible indications for performing a given procedure 
but only those which generally are agreed upon. 
Those criteria which were thought to be controversial 
or subject to inappropriate application were not 
included; it was thought that these cases should be 



subjected to peer review by the physician advisor. 
The criteria sets are in no sense an attempt to define 
standards of quality or to preclude innovation by 
physicians. The final sample criteria sets will be 
distributed in 1980. 

Evaluation of Surgical Utilization 

A contractor was retained to analyze the use of 
surgical services by Medicare and Medicaid patients 
in selected PSRO areas and to address the impact 
of PSRO review on aggregate and procedure-specific 
surgical discharge rates. The study includes extensive 
analysis of small area variations in surgical rates 
for the selected PSRO areas (California V, Montana, 
Wisconsin I, New Jersey I, and Washington, D. C.) 
and the analysis of changes in surgical practice 
patterns over the period 1977-1980. The contractor 
is developing a technical assistance manual on 
analysis of surgical services utilization. 



8.3.6 Other Forms of Review 

8.3.6.1 LONG-TERM CARE REVIEW 

PSROs are required to review care in all skilled 
nursing facilities. Additionally, a PSRO may perform 
review in intermediate care facilities and those for 
the mentally retarded (ICF-MRs), if requested by a 
State Medicaid agency or if the Secretary finds that 
the State is not performing effective review or that 
the State is not reviewing the combined SNF/ICFs 
efficiently. 

HSQB has developed model guidelines for long-term 
care (LTC) review which may include elements 
addressing preadmission certification, concurrent 
review, medical care evaluation studies, and profile 
analysis. PSROs have been encouraged to develop 
creative review approaches, based on the model 
guidelines and considerations of the characteristics 
of the facilities, patients, and local health care 
practitioners. 

The following are priority goals for guiding PSROs 
in their long-term care review functions: 

1. To assure that admissions to skilled nursing 
facilities and intermediate care facilities and 
ICF-MRs (if applicable) are appropriate. The 
nature of the long-term care patient population 
and limited experience to date suggests that 
efforts be placed on assuring appropriate 
admissions and close evaluation early in the 
patient's stay in a long-term care facility rather 
than focusing efforts on continued stay review 
and, in particular, on patients who have been 

in the facility six months or longer. 

2. To improve the quality of care provided to 
patients in long-term care facilities through the 
review program which incorporates a coopera- 
tive working relationship with practitioners, 
facilities, and other responsible agencies and 
organizations. 



174 



3. To decrease barriers to the appropriate 
placement and ongoing care of the long-term 
care patient through cooperative efforts with 
community and state agencies and organizations. 

4. To conduct effective review at a reasonable 
cost. 

In LTC review, the patients' durations of stay allow 
inpatient review to be performed, and such review 
is expected to be more effective than in short-stay 
hospitals. However, many features of the LTC system 
are observed to present practical and theoretical 
difficulties which hamper the use of review 
methodologies previously developed for acute care 
hospital review." Those features are: 

1. More LTC facilities than acute care hospitals 
exist in a given PSRO region (nationally a 
four-fold difference). PSRO delegated reviews 
are not permitted in LTC facilities. Thus, a LTC 
Review System must enter into more institutions 
and interface with more geographically-dispersed 
administrative entities. 

2. Compared to hospitals, LTC facilities are often 
smaller, have lower staff-patient ratios, higher 
personnel turnover, and lack the technological 
or research base to facilitate PSRO activities. 

3. Medical records in LTC facilities often tend to 
be too elementary and inadequate to provide 
information for review. 

4. Physicians are less involved in patient 
management in LTC facilities. This factor, 
together with the indeterminate bounds of 
complete physical and/or psychosocial care, 
cause LTC to be delivered mostly by non- 
physician personnel. To ensure greater effective- 
ness, more non-physician inputs are needed 

in the design of LTC review activities. 

5. Characteristics of LTC patients create problems 
in the establishment of review criteria. For 
instance, patients have multiple diagnoses, need 
psychosocial care in addition to medical care, 
and are often not expected to recover fully 
and/or return to the community. Hence, criteria 
to measure the psychosocial well-being and 
definitions of quality of care tend to be elusive. 

6. Reviews to reduce overutilization and mandates 
to reduce occupancy rates may threaten the 
viability of LTC facilities, which are generally 
much smaller in scale than hospitals. 

In view of these differences, HSQB plans to revise 
the existing LTC review guidelines, which are 
modeled after the acute hospital review system. 

Of the 15 demonstrations initiated in FY 1977 to 
test various approaches to LTC review, five were 



excluded from the original evaluation plan and ten 
projects ^2 were assessed '^ in FY 1979. 

The Rand evaluation concluded that the PSROs 
can develop monitoring programs with diverse 
approaches to LTC review, that these programs can 
assure better quality of care, but that they will not 
reduce the number of dollars spent on long-term care. 
Rand also found that because the demonstration 
projects had not been mandated to evaluate the 
effectiveness of their review, they did not collect 
baseline data, establish control groups, or make 
other provisions for measuring the impact of their 
projects. 

Concerning specific types of review. Rand concluded 
that preadmission review can reduce the days of 
care under Medicare and may improve the transfer 
of information from hospital to nursing home. 
Admission review, which takes place after a patient 
has been transferred from the hospital to the nursing 
home, is considered to be more expensive and 
disruptive than preadmission review. Continued stay 
review is felt to be effective for nursing home patients 
only in the first six months of stay, since patients 
tend to become institutionalized after that point and 
denial rates tend to decline. Concurrent quality 
assurance requires professional judgment to identify 
quality problems; such problems can be detected 
in the patient's chart, but bedside review enhances 
the process. Rand recommended multi-disciplinary 
involvement in developing a review process based 
on more explicit criteria, psychosocial outcome review, 
epidemiological studies, drug review, and develop- 
ment of data systems that can document the PSRO's 
achievements. 

There are currently 55 PSROs funded for LTC 
review. Of these, 43 are due for refunding in FY 1980, 
with an $11 million national budget. The remaining 
12 will not be refunded until FY 1981 since they had 
received funding in FY 1979 for more than twelve 
months. Many other PSROs intend to perform LTC 
review, but tlieir initiation or expansion in this area 
has been prevented bv the FY 1980 budget cut. 
Currently, about 26 PSROs are performing ICF review 
and 9 PRSOs are performing ICF-MR review in 
LTC facilities. 

8.3.6.2 AMBULATORY CARE REVIEW 

The Medicare-Medicaid Anti-Fraud and Abuse 
Amendments (P.L. 95-142) require PSROs to begin 
reviewing ambulatory care not later than two years 
after their full designation. It also requires that PSROs 
requesting funds for the review of care in shared 
health facilities be given priority. 



" Ten Demonstration Projects in PSRO Long-Term Care 
Review. Vol. 1. Rand Corporation, May 1979. 



^ California VIM, Colorado, Maryland II, Massachusetts III, 
Minnesota I, New York XVI, Oregon, South Carolina, Utah, 
and Vermont. 

" Rand Corporation, op. cit., Volumes 1 and 2. 



175 



The long-range goal of PSRO activities has been 
to develop cost-effective ambulatory review systems 
which will serve as models for implementation 
throughout PSROs. In keeping with this goal, five 
PSROs (Massachusetts Area II, New Mexico, New 
York Area XVI, Oregon Area II and Pennsylvania) 
were funded as demonstrations in 1976. In 1979 all of 
the projects were completed. The major finding 
from the demonstrations i* was that an approach based 
on claims data has potential for cost-effectiveness. 
More demonstrations, however, would be required 
prior to adopting any particular approach as a 
prototype for all PSROs. 

Currently six demonstration projects are funded. 
Three projects. New Mexico PSRO, Utah PSRO, and 
National Capitol PSRO, have developed approaches 
based on a review of Medicaid claims data. National 
Capitol also reviews Medicare claims data. Two 
New York PSROs, the Bronx and New York County, 
are performing on-site audits of shared health 
facilities. Dade-Monroe PSRO in Florida performs an 
office audit of shared health facilities cited for further 
study following a review of Medicare claims data. 

In FY 1979, special initiative funds were allotted 
to Maine Area I and the District of Columbia. During 
the period, the National Capital Medical Foundation 
in the District of Columbia implemented a program 
to encourage performance of selected surgical 
procedures on an ambulatory basis in hospitals. While 
the program began as a study effort, it became 
mandatory, and PSRO findings relating to therapeutic 
abortions, dilation and curettage, and non-newborn 
circumcision are binding for payment purposes. 
During the study period, a 50 percent reduction in 
admission for the procedures was observed. ^^ 
The PSRO states that this special review program 
has contribtued to improved quality of care. 

Based on the preliminary stage of PSRO program 
ambulatory care review development, current agency 
directions are: 

1. To continue funding projects as demonstrations. 

2. To encourage review concepts based on 
existing data bases. 

3. To encourage review concepts which have 
potential as prototypes for universal imple- 
mentation. 

4. To encourage additional cost-effective ap- 
proaches to review. 

8.3.6.3 ANCILLARY SERVICE REVIEW 

PSROs have a statutory responsibility to review 
not only hospitalization, but also the ancillary services 
rendered during hospitalization. An ancillary service 
may be defined as any service incident to hospitaliza- 



tion, other than room and board, nursing, dietary, 
and physician services. Examples of the category 
are laboratory, pharmacy, radiology, and occupational, 
physical, respiratory, and speech therapy. 

Inappropriate use of ancillary services can subject 
patients to significant unnecessary risk, and these 
services may account for over 50 percent of hospital 
charges on a national basis. Therefore, HSQB expects 
that PSROs, through ancillary service review, can 
improve quality of care while reducing health care 
costs. 

Ancillary Service Review (ASR) may be implemented 
as a part of the hospital review system, using similar 
general approaches and the same data base, or as a 
separate review system using a variety of data bases. 
The major review components presently available 
to PSROs are: (1) concurrent review, including 
preprocedure review, focused concurrent review, 
concurrent quality assurance, etc.; and (2) retro- 
spective review, whether by specific MCE studies or 
profile analysis. 

In FY 1978, a total of 10 PSROs — Arkansas, 
California Area XXII, Connecticut Area I and Area II, 
Maryland Area VI and Area VII, New York Area IX 
and Area XV, Rhode Island, and Wyoming — were 
each funded with $30,000 for demonstration projects. 
The emphasis was to develop cost-effective systems 
utilizing existing review methods and existing data. 
The results of the demonstration projects were 
inconclusive. 

During FY 1979, 62 more advanced and innovative 
PSROs were funded to develop ASR systems. The 
actual funding was $1,980,080 and represented 
about 75 percent of the special initiative funding. 
Resources were allocated as follows: 



HEW Region '« 


PSRO Areas 
2 


Funding 


1 (Boston) 


$ 147,000 


II (New York) 


11 


280,486 


III (Philadelphia) 


8 


160,987 


IV (Atlanta) 


13 


515,942 


V (Chicago) 


13 


411,443 


VI (Dallas) 








VII (Kansas City, MO) 


4 


140,080 


^111 (Denver) 


2 


510,000 


IX (San Francisco) 


8 


227,605 


X (Seattle) 


1 


45,537 
$1,980,080 



" Health Care Management Systems, Cooperative 
Demonstration and Analysis of Ambulatory Care Quality 
Assurance Review Methodology in PSRO Settings, 1978. 

"* National Professionals Standards and Review Council 
Meeting, November 5-6, 1979. 



HSQB serves as a source of program guidance 
in ASR to PSROs. It monitors the special initiatives 
by assisting Regional Offices in evaluating the PSROs' 
ASR proposals and maintaining an ASR tracking 
system. X-ray referral criteria developed through 
FDA's Bureau of Radiological Health are sent to 
PSROs as they are developed and validated. Through 
two expert panels on respiratory therapy, problems 
were defined and review criteria developed. In FY 
1980, HSQB will refine and develop ASR 



' Headquarters city is shown for each region. 



176 



methodologies. A general system for drug prescribing 
review will be designed and suggested to PSROs. 

Besides the special initiative funding, other PSROs 
are also conducting ASP within their present budgetary 
constraints; e.g., through MCE studies. Tables 100 
through 103 reflect data obtained from a review of 
PSROs' current grant applications and contract 
proposals. The data reflect ASRs conducted during 
the concurrent performance period, and may include 
some of those PSROs receiving special initiatives 
funding. As cautioned before, the reader should not 
interpret the data to mean that the remaining PSROs 
did not conduct ASR; data were not available from 
the funding documents for PSROs other than those 
covered here. Also, the data reported did not include 
the number of reviews conducted or the number of 
patients involved. A review of the available data 
revealed that: 

• A total of 154 forms of ASR were reported by 
35 PSROs in 8 regions. 

• Among the forms employed to conduct ASR, 
a majority (59.7%) reported utilizing medical 
care evaluations. 

• Of the 154 forms of reviews reported, 61 or 
39.6 percent were conducted on an areawide 
basis rather than in selected hospitals. 

No review was indicated in Region VI and VII 
funding documents.!^ 



'Again, it is empiiasized tliat tiie absence of an explicit 
reference to ASR in a grant or contract document does not 
necessarily denote that no sucti review is being performed. 



8.3.6.4 PHYSICIAN SERVICES REVIEW 

Under hospital review, PSROs are responsible for 
reviewing physician services through concurrent 
review, medical care evaluation studies, and profile 
analysis. However, without coordination with the 
carriers, PSROs have not been able to impact on 
payments to physicians for services identified as 
medically unnecessary or inappropriate in the hospital 
setting. In FY 1979, PSROs and carriers were 
directed to work together to develop an interface 
for the management of the denials of Part A, the 
hospital insurance, and Part B, the medical insurance, 
of the Medicare program. The approach is to 
complement the existing hospital review activities. 
The intent is to design a review system to monitor 
patterns of care, focus detailed review on problem 
areas and, when appropriate, make adverse review 
determinations for claim payment denial. The PSRO 
will have to access the carrier's physician profiles, in 
order to identify or refine its physician profiles. 

This approach to physician services review 
conceptually satisfies the PSRO legislative responsi- 
bility. The objective in physician services review 
is to capitalize on the professional medical expertise 
of the PSROs and the data collection and manipulation 
capabilities presented by the carriers' data systems. 
Through this mechanism, the PSRO and carrier 
will strengthen the effectiveness of medical necessity 
review of physician services and reduce duplication 
and overlap in data collection and review activities. 
This represents a new but important relationship 
for PSROs and carriers. 



TABLE 100 
Summary of Ancillary Service Reviews 



Region 



Number of 
PSROs 

Conducting 
Hospital 
Review 



Number of 

PSROs 

Conducting 

Ancillary 

Service 

Reviews 



Number of 

Ancillary 

Service 

Procedures 

Being 
Reviewed 



Average 

Number of 

Ancillary 

Service 

Procedures 

Reviewed 

Per PSRO 



I 

II 

III 

IV 

V 

VI 

VII 

VIII 

IX 

X 



Total 



13 
26 

27 

24 

41 

7 

7 

6 

31 

5 

187 



5 
5 

10 
2 
5 


3 
3 

_2 

35 



19 
27 
48 

6 
20 




12 
15 

7 

154 



3.8 
5.4 
4.8 
3.0 
4.0 


4.0 
5.0 
3^ 

4.4 



Source: Grant Applications and Contract Proposals. 



177 



Service 

Category | 



TABLE 101 
Summary of Ancillary Service Review Activity, by Region 

Number Conducted by Region 



III 



IV 



VI 



VII 



VIII 



IX 



Total 



Pet. of 

All 
Reviews 



Blood 
Diagnostic 

X-Ray 
Drugs 
Laboratory 
Occupational 

Therapy 
Physical 

Therapy 
Respiratory 

Therapy 
Other 



1 



20 



13.0 



3 
4 
4 

4 

4 

3 
3 



2 
2 

1 


2 

1 
1 


1 
1 
1 


19 
20 
19 


12.3 
13.0 
12.3 


1 








8 


5.2 


1 


3 





17 


11.0 


2 
2 


• 1 
5 


1 
1 


23 
28 


14.9 
18.2 



Totals 



19 



27 



48 



6 



20 



12 



15 



154 



100.0 



Source: Grant Applications and Contract Proposals. 



TABLE 102 
Summary of Ancillary Service Review Activity, by Methodology 





Number Conducted, 


by Review Component 




Number 


Service Category 


Concurrent 


MCE 


Other 


Total 


Conducted 




Review 


Study 


Area-Wide 


Blood 


1 


13 


6 


20 


7 


Diagnostic X-Ray 


1 


11 


7 


19 


6 


Drugs 


3 


14 


3 


20 


7 


Laboratory 


3 


10 


6 


19 


9 


Occupational Therapy 


2 


5 


1 


8 


5 


Physical Therapy 


3 


9 


5 


17 


10 


Respiratory Therapy 


3 


16 


4 


23 


9 


Other 


4 


14 


10 


28 


8 


Totals 


20 


92 


42 


154 


61 



Source: Grant Applications and Contract Proposals. 



HSQB plans to implement Physician Services 
Review on a phased basis. In order to obtain practical 
knowledge about the possibilities of cooperative 
efforts between local PSROs and Medicare carriers, 
five initial projects are planned (see Table 104). 

8.4 Analysis of 1978 PSRO Performance 
Ratings 

8.4.1 Background 

In 1978 HSQB staff designed an instrument to 
collect data that would contribute to an assessment 
of PSRO performance and organizational effectiveness. 



The instrument was to be completed by regional 
project officers on all PSROs which had conducted 
hospital review in all or a portion of calendar year 

1 977. The PSROs were to be rated by the project 
officers on the basis of PSRO performance for the 
period March 1977 through August 1978. In August 

1978, 38 project officers rated 109 PSROs. 

The instrument consisted of over 50 items which 
were grouped into the categories of physician 
involvement in PSROs, general management, external 
relationships, implementation of review, medical care 
evaluations, sanctions and denials, other hospital 
review being performed, and responsiveness in the 
review process. 



178 



TABLE 103 
Summary of Ancillary Service Review Activity, by Region and Methodology 



Region 



Concurrent 
Review 



Methodology 



MCE 
Study 



Otiier 



Total 



III 

IV 

V 

VI 

VII 

VIII 

IX 

X 



17 

15 

35 



12 





3 

3 

7 



2 
8 
11 
6 
6 


1 
8 




19 
27 
48 

6 
20 




12 
15 

7 



Totals 



20 



92 



42 



154 



Pet. of All Reviews 



13.0 



59.7 



27.3 



100.0 



Source: Grant Applications and Contract Proposals. 



TABLE 104 
PSROs Funded for Physician Services Review 



PSRO Area 



Funds ($)' 



Massachusetts Area I 
Montana 

Washington State 
Pennsylvania Area VIII 
California Area XXIII 



$23,396 
20,000 
30,000 

No additional 
15,000 (estimate) 



* Fundings for all demonstration projects, except those in 
Massachusetts Area I and Pennsylvania Area VIII, are for 
18 months. These two PSROs are scheduled for budget 
negotiation in the first quarter of FY 1980; this may affect 
their budgets for Physician Services Review. All of these 
projects are expected to be completed in 1981. 

The 1978 PSRO evaluation presented a preliminary 
analysis of the August 1978 project officer ratings. 
The analysis focused primarily upon the presentation 
of frequency distributions, average scores, and 
correlations between survey components and total 
scores. Data availability and time constraints limited 
the analyses. 

8.4.2 Objectives 

The analyses presented last year were limited to: 

1 . Averages and ranges of total scores and 
subscores. 

2. Correlations between subscores and total score. 

3. Correlations between subscores and total 
scores by PSRO age. 

This year's effort seeks to expand upon these 
analyses and provide more in-depth analyses for the 
purpose of providing a broader description of the 



operation and management of PSROs as of 1978 and 
to examine the relationship between PSRO manage- 
ment and review processes and hospital utilization 
and cost. 

The objectives of this year's expanded study are 
as follows: 

1. Present additional analysis of specific items, 
components, and their interrelationships. 

2. Develop a measure of organizational performance 
based upon key interrelated items in the 
questionnaire that will serve as an indicator 

of overall PSRO effectiveness. 

3. Test a series of study hypotheses to determine 
the relationship of specific items to: (a) the 
new measure of overall PSRO performance 
(2 above), and (b) measures of hospital 
utilization and cost. 

8.4.3 Methodology 

A three-fold approach was employed to achieve 
the objectives: 

1. An analysis was performed of the response 
distribution, general categories, and interrela- 
tionship of specific questions. 

2. A factor analysis was performed to determine 
a set of interrelated items that can be used as 
an improved measure ("score") of overall 
PSRO performance. 

3. A set of study hypotheses was developed that 
seeks to measure the relationship of selected 
process and utilization factors to the new PSRO 
performance score. The hypotheses and methods 
utilized were: 

a. High scores on questions related to 
innovation will be associated with overall 
high ratings. 



179 



The sum of the rating scores on seven 
items relating to innovation were correlated 
against the total performance score. The 
seven items were: 

1. Use of data to draw conclusions 
about utilization. 

2. Identification of specific utilization 
problems. 

3. Development of action plans or 
strategies to impact on identification 
problems. 

4. Ability to document that its actions 
have impacted on problem areas. 

5. Willingness to reduce concurrent 
review. 

6. Ability to move rapidly towards 
exempting a significant number of 
patients from review. 

7. Informs providers of their 
performance. 

b. PSROs which are willing to be aggressive 
in monitoring hospitals and initiating 
adverse actions, where appropriate, will 
have high overall scores. 

The sum of the scores on six items relating 
to aggressiveness were correlated against 
the total performance score. The six 
items were: 

1. Ongoing monitoring of delegated 
hospital review. 

2. Withdrawal of delegation from 
hospitals under review. 

3. Number of denials issued in the 
past 18 months. 

4. Proceedings for potential sanctions 
initiated against a provider. 

5. Actual recommendation for sanction 
forwarded to Secretary. 

6. State agency/DHEW notified of 
potential fraud or program abuse. 

c. Dynamic PSROs will tend to have higher 
total scores than other PSROs. 
Sixteen items comprised the dynamic 
scale and were correlated against the 
total performance score. The scale included 
the seven items related to innovation, 

the six items related to aggressiveness, 
and the following three items: 

1. Level of physician participation 
in PSRO review activities. 

2. Competence of executive leadership. 

3. Interactions with health system 
agencies. 

d. The use of data systems is related to 
overall j^erformance. 

The sun of the scores on five items 
relating to the use of data systems were 
correlated against the total performance 
score. The items were: 
1. Data system processsing capability. 



2. Timeliness and completeness of 
reports. 

3. Validation system for Inputs. 

4. Use of data inputs in review system. 

5. Use of data to draw conclusions about 
utilization problems. 

e. The level of PSRO/physician and 
PSRO/hospital interaction is related to 
overall performance. 
The sum of the scores on five items 
relating to physician and hospital 
interactions were correlated against the 
total performance score. The items were: 

1. Level of actual physician participation 
in PSRO review activities. 

2. Ongoing monitoring of delegated 
hospital review. 

3. General relationships with hospital. 

4. Withdrawal of delegation from 
hospitals. 

5. Training provided for hospitals, 
review coordinators, and physician 
advisors. 

8.4.4 Data Limitations 

The data obtained in the 1978 project officer 
performance ratings should be understood within 
the context of the following considerations: 

1. The project officer ratings were the only data 
available for obtaining information on the 
organizational aspects of conditional PSROs. 

2. The performance ratings may be affected by 
project officer rating biases, such as: 

a. Differing levels of experience with their 
PSROs at the time ratings were done, 

b. Inclinations towards a particular PSRO or 
to ratings in general; this is referred to 
as a "halo" effect (i.e., where a rater's 
overall assessment will color all responses 
either in a positive or negative direction), 
and 

c. Time availability for completing the 
assessment. 

3. Varying response categories in many questions 
made it difficult to compare some items with 
others and to construct scales for measuring 
different organizational aspects of PSRO 
performance. 

4. The instrument was not pretested, resulting in: 

a. A lack of discrimination between PSROs 
on some items (i.e., raters accord the 
same response category to all PSROs), 
and 

b. A lack of understanding for some of the 
questions. 

5. A lack of information by project officers on 
some items led to high non-response rates. 
Specifically there are six such items (four 
related to the MCE component): 

a. Medicaid monitoring — 66.1 percent non- 
response. 



180 



b. Deficiencies eliminated in MCE reaudits — 
54.1 percent non-response. 

c. Required reaudits performed — 34.9 
percent non-response. 

d. Required MCEs performed — 16.5 percent 
non-response. 

e. Appropriateness of MCE topic selection — 
12.8 percent non-response. 

f. PSRO hospital review budget — 10.1 percent 
non-response. 

Because of these limitations, the information 
presented must be viewed as relative indicators of 
performance rather than absolute scores of 
performance. 

8.4.5 Findings 

8.4.5.1 GENERAL 

A tabulation of all questions revealed high scores 
(above the mid-point value) by project officers on 
the following 10 items: 

1. Physician participation in PSRO review 
activities. 

2. Stability of governing body. 

3. Management of fiscal affairs. 

4. Level of agreement between Medicare and 
PSRO decisions. 

5. Level of agreement between Medicaid and 
PSRO decisions. 

6. Implementation of concurrent hospital review. 

7. Monitoring of delegated hospital review. 

8. Training provided. 

9. Identification of specific utilization problems. 
10. PSROs informing providers of their performance. 

Low scores (below the mid-point value) were 
recorded for the following eight items: 

1. Number of PSROs which have withdrawn 
delegation from hospitals. 



2. The number of areawide medical care 
evaluation studies performed. 

3. Ancillary services review. 

4. Review of physician services. 

5. Proceedings for potential sanctions initiated. 

6. Sanctions forwarded to the Secretary. 

7. State agency notified of fraud or abuse. 

8. PSRO will be able to exempt patients from 
review. 

A separate analysis was performed on the eight 
general categories of items in the questionnaire: 
physician involvement, general management, external 
relationships, implementation of concurrent review, 
medical care evaluations, other hospital review 
activities, responsiveness in review, and use of 
sanctions, and denials, and although the limitations 
of the data prevented computation for individual 
PSROs, the eight general components were tabulated 
for all PSROs. Table 105 shows the potential and 
actual scores for each of the components. 

Table 105 reflects, by questionnaire component that: 

1. PSROs score high (above the mid-point scale 
value) for: physician involvement, general 
management, external relations, concurrent 

review, and responsiveness in review. 

2. PSROs score low (below the mid-point scale 
value) for: medical care evaluation studies, 
other forms of hospital review (i.e., other than 
concurrent), and sanctions and denials. 

Individual items were also analyzed in order to 
gain a better understanding and perspective on 
PSRO performance and organizational effectiveness 
in 1978. 

A. Physician memberstiip and participation 

The first two questions on the survey instrument 
examined physician membership and physician 
participation in PSROs. 



TABLE 105 

Ranges and Average Scores for all PSROs (109) 
for Eight Questionnaire Components 











Percent of 


Component , 


Possible Range 
of Scores 


Actual Range 
of Scores 


Average 
Score 


Possible 
Range 


Physician Involvement 


6-28 


9-27 


19.9 


71.1 


General Management 


10-50 


18-50 


33.4 


66.8 


External Relations 


4-20 


10-20 


14.7 


73.5 


Concurrent Review 


6-27 


11-26 


19.3 


71.2 


Medical Care 


5-25 


7-22 


14.6 


58.4 


Evaluation Studies 










Other Review 


3-13 


3-13 


5.5 


42.3 


Sanctions 


4-14 


4-11 


7.0 


50.0 


Responsiveness in 










Review 


9-30 


15-28 


20.7 


69.0 


Total Score 


47-207 


77-197 


135.1 


65.3 



181 



Physician membersfiip in PSROs was rated as 
being very higii as reflected in Table 106. 



TABLE 106 

Distribution of PSROs by Percentage of 
Area Physicians Belonging to PSRO 



Percentage of Physicians 


Percentage 


Belonging to PSRO 


Distribution 


25-35% 


1% 


36-50% 


18% 


51-65% 


43% 


66-75% 


28% 


over 76% 


10% 



Although these membership levels appear to be 
high, an additional measure of physician involvement 
is the percentage of physicians participating in PSRO 
activities. Project officer ratings of physician 
participation are presented in Table 107. 



TABLE 107 

Distribution of PSROs by Level 
of Physician Involvement 



Level of Physician Involvement 



Percentage 
Distribution 



Non-existent 2% 

1- 5% of the membership 15% 

6-10% of the membership 30% 

11-20% of the membership 33% 

greater than 20% of the membership 20% 

When membership levels are compared to actual 
physician participation, the relationship appears to 
be weak with a correlation coefficient of .29. This 
suggests that although a PSRO might have a high 
level of membership, it does not necessarily indicate 
a correspondingly high level of physician participation. 
The cross tabulation in Table 108 demonstrates that 
the average PSRO has a membership level of about 
one-half to three-quarters of the area's physicians, 
but that only 6 to 20 percent of them actively 
participate in PSRO activities. 



TABLE 108 

Level of Physician Participation by Level 
of Physician Membership in PSRO 



Level of Physician 
Participation 



Level of Physician Membership 



25-35% 



36-50% 



51-65% 



66-75% 



76-100% 



Non-Existent 

1- 5% 

6-10% 
11-20% 
Greater than 20% 



1% 



1% 


— 


— 


— 


7% 


4% 


3% 


1% 


5% 


16% 


8% 


1% 


1% 


15% 


10% 


7% 


4% 


8% 


7% 


1% 



B. Staff Competency and Stability 

A key concern in PSRO organizations is the 
competence of the leadership and staff. Overall, 
project officers rated the staff highly as illustrated 
in Table 109. 

Since it is important that PSROs be able to retain 
competent staff over a sufficiently long period to 
achieve any impact, the PSROs' ability to retain them 
was examined by correlating executive and 
administrative competency levels with staff stability. 
The correlation coefficients were .43 and .41, 
respectively, suggesting a comparable ability to retain 
competent executive leadership and administrative 
staff. 



TABLE 109 

Ratings of Competency of PSRO Executive 
Leadership and Administrative Staff 



Competency of 


Competency of 


Executive 


Administrative 


Leadership 


Staff 


Excellent 25% 


Excellent 16% 


Good 37% 


Good 49% 


Average 22% 


Average 24% 


Fair 12% 


Fair 10% 


Poor 3% 


Poor 0% 


Unknown 1 % 


Unknown 1 % 



182 



Tables 110 and 111 show the relationship between 
the stability and competency for each group. It 
should be noted that when considering executive 
directors, the most frequent combinations are 
executive directors who are rated as either good or 



excellent leaders with good or excellent stability. 
The administrative staff, on the other hand, were 
more likely to be rated good with only average or 
good level of stability. 



TABLE 110 
Competency Compared to Stability of Executive Directors 



Stability 



Competence of Executive Directors 



Unknown 



Poor 



Fair 



Average 



Good 



Excellent 



Unknown 

Poor 

Fair 

Average 

Good 

Excellent 



1% 



1% 



1% 



1% 


2% 


— 


— 




— 


1% 


4% 


1% 


1% 


1% 


6% 


5% 


10% 


1% 


1% 


1% 


8% 


22% 


8% 


- — 


2% 


4% 


4% 


14% 



TABLE 111 
Competency Compared to Stability of Administrative Staff 



Stability 



Competence of Administrative Staff 



Unknown 



Poor 



Fair 



Average 



Good 



Excellent 



Unknown 

Poor 

Fair 

Average 

Good 

Excellent 



1% 



— 


1% 


— 


1% 


1% 


2% 


— 




2% 


3% 


3% 




3% 


7% 


13% 




2% 


10% 


23% 


5% 


2% 


1% 


10% 


10% 



C. Making and Meeting Goals and Objectives 

PSRO ability to set and meet objectives has been 
established as a management initiative. In 1978 
project officers rated the majority of conditional 
PSROs as doing average or better in this area. The 
ratings of PSROs setting of goals is shown in Table 
112. 

A PSRO's success in meeting goals appeared to 
be related to their content or substance with a 
correlation of .66 between these two items. Table 113 
illustrates this relationship. 



TABLE 112 

Rating of PSRO Goal Setting and Success 
in Meeting Goals 



Goal and Objective 


Success 


in Meeting 


Setting 




Goals 




Poor 


6% 


Poor 




12% 


Fair 


17% 


Fair 




14% 


Average 


35% 


Average 




28% 


Good 


32% 


Good 




42% 


Excellent 


9% 


Excellent 




3% 


Unknown 


2% 


Unknown 




2% 



183 



TABLE 113 
Setting of Goals and Objectives Related to Success in Meeting Them 



Success in 

Meeting Goals 

& Objectives 




Content 


or Substance of Objectives 


and Goals 




Unknown 


Poor 


Fair 


Average 


Good 


Excellent 


Unknown 








1% 


1% 








Poor 


1% 


5% 


4% 


1% 


1% 


— 


Fair 


1% 


1% 


5% 


7% 


— 


— 


Average 


— 


— 


5% 


13% 


8% 


2% 


Good 


— 


— 


2% 


13% 


22% 


5% 


Excellent 


— 


— 


— 


— 


1% 


2% 



D. Selected PSRO Descriptions 

A descriptive analysis of organization effectiveness 
was conducted that considered the interaction 



among selected factors in three PSROs — one ranking 
in the top 10 percent, one in the middle range, and 
one in the lower 10 percent. Table 114 shows the 
variations among these factors for the three PSROs. 



TABLE 114 
Description of Three Selected PSROs 



Factors 


Good PSRO 


Average PSRO 


Poor PSRO 


(In Top 10%) 


(In Middle Range) 


(In Bottom 10%) 


Leadership 








• governing body 


excellent 


excellent 


lacking 


• executive 


excellent 


excellent 


fair 


• administrative staff 


excellent 


excellent 


fair 


Committees 








• policy decisions 


quite active 


average 


little activity 


• physician participation 


average 


average 


virtually non-existent 


Goals and Objectives 








• statement of by PSRO 


excellent 


good 


fair 


• success in meeting 


good 


good 


fair 


Data 








• validation system 


good 


average 


not able to rate 


• development of criteria 


good 


average 


not able to rate 


• use of data 


excellent 


average 


not able to rate 


Relationships 








• with hospitals 


good 


good 


good 


• training 


excellent 


incomplete 


incomplete 


Problem Impact 








• development of plans 


average 


begun to develop 


begun to develop 


• strategies 


average 


begun to develop 


begun to develop 


• documentation 


good 


not yet able 


begun to develop 



8.4.5.2 ORGANIZATIONAL PERFORMANCE 

To arrive at a composite measure of PSRO 
organizational performance a factor analysis was 
conducted on all survey items that met the following 
criteria: 

1. Understood by project officers. 

2. Answered by most project officers. 

3. Reflected variation of response. 



4. Felt to be important indicators by project 
officers. 

The result of the factor analysis resulted in an 
indicator of performance that was representative 
of most of the instrument and included the following 
15 items: 

1. Management capability, initiative, and leadership 
exercised by the governing body. 



184 



2. Level of committees in PSROs in policy-making 
and activities. 

3. Level of physician participation in committees. 

4. Competence of executive leadership. 

5. Competence of administrative staff. 

6. Content or substance of objectives and goals. 

7. Past success in meeting organizational goals 
and objectives. 

8. Validation system for data inputs — level and 
extent of monitoring and data validation 
activity. 

9. Use of data outputs in review system. 

10. Development/adoption of acceptable, specific 
criteria and standards, ongoing review and 
revision. 

11. General relationships with hospital (level of 
cooperation). 

12. Training provided for hospitals, review coordi- 
nators, and physician advisors. 

13. Use of data to draw conclusions nbou'. 
utilization problems/patterns of care. 

14. Development of action plans or strategies to 
impact on identified problems. 

15. Ability to document that actions have impacted 
on problem areas. 

The 15 items that comprise this measure of 
organizational effectiveness were adjusted to a 100 
point scale. PSRO scores ranged from a high of 97.1 
points to a low of 34.3 points, with an average of 
67.8 points. Average scores for each of the 10 regions 
are displayed in Table 115. 



TABLE 115 

Average Organizational Effectiveness Scores 
by HEW Region 



Region 



States and Territories 



Points 



I MA, CN, Rl, NH, VT, ME 71.1 

II NY, NJ, VI, PR 65.8 

III PA, MD, DL, VA, WV 70.2 

IV GA, FL, MS, AL, NC, SC, TN, KY 76.9 

V IL, IN, OH, Ml, MN, Wl 67.4 

VI TX, NM, OK, AK, LA 62.9 

VII MO, KS, NB, ID 63.9 
VIM CO, UT, WY, Ml, ND, SD 69.0 

IX CA, NV, AR, HI 66.0 

X WA, OR, ID, AK 58.0 



Total 



67.8 



8.5 PSRO Performance and Utilization 



The mean score for all PSROs was 16.7 points. 
PSROs could achieve a maximum of 24 points. 
Regions VI and VIII had the highest scores; 
Regions VII, IX, and X the lowest. When 
correlated, the coefficient between the two sets 
was .72 indicating that those PSROs which are 
innovative are also performing well. When 
overlapping items were excluded, the correlation 
remained statistically significant at .50. 
PSROs which are willing to be aggressive in 
monitoring hospitals and initiating adverse 
actions, where appropriate, will have high 
overall scores. 

The mean score for all PSROs was 11.9 points 
out of a possible 22 points. Regions I, IV, and 

IX had the highest scores; Region VII, VIII, and 

X had the lowest. The correlation coefficient 
was statistically significant (.32) suggesting that 
aggressiveness is related to other positive 
organizational characteristics. When the over- 
lapping adjustment was made, the relationship 
remained significant at .69. 

Dynamic PSROs will tend to have higher total 
scores than other PSROs. 

The mean score was 39.4 points out of a possible 
61 points. PSRO in Region I, IV, and VIII, had 
the highest scores with Regions II, VII and X 
scoring the lowest. The correlation coefficient 
(.84) suggests that commitment to action is 
related to other positive organizational features. 
The relationship may, in some part, be due 
to an overlap of four items on the two scales. 
The use of data systems is related to overall 
performance. 

The mean score was 14.7 points out of a 
potential of 24 points. Regions I, IV, and VI 
recorded the highest scores with Regions 
VII, VIII, and IX recording the lowest. 
The correlation coefficient (.72) suggests that a 
relationship exists between performance in the 
data systems component and overall per- 
formance; some of this may be related to an 
overlap in the performance measures. However, 
with the exclusion of these items, the correlation 
remained statistically significant at .54. 
The level of PSRO/physician and PSRO/hospital 
interaction is related to overall performance. 
PSROs recorded a mean of 18.2 points. A 
maximum of 25 points were possible. Regions 
IV, VI, and IX had the highest scores and 
Regions II, VII, and X the lowest. As above, 
with exclusion adjustments significance remained 
at .54. 



Analyses were conducted to test the hypotheses 
presented in 8.4.3. The findings, keyed to the original 
hypotheses, are as follows: 

a. High scores on questions related to innovation 
will be associated with overall high ratings. 



8.6 Conclusion 

The information presented in this chapter indicates 
that the PSRO program is in a transitional stage. 
Programmatically, the scope of review activities has 



185 



broadened to encompass other areas of health 
services besides basic hospital review, Also, essential 
program activities are evolving through policy, 
administrative, and methodological refinement. In 
particular, concurrent review is becoming more 
focused; the essential concepts of medical care 
evaluation are undergoing evolutionary change; and 
profile analysis is becoming more widely utilized. 
Having developed the "process" thus far, the 
program is under increasing Congressional pressure 
to demonstrate "impacts" with regard to quality. 



utilization, and cost of health care. This, together with 
the limitation in funding levels, puts the program 
under pressure to do more and perform better with 
constrained resources. 

Consequently, general management initiatives in 
objective setting, self-assessment, financial manage- 
ment, and relationships with external agencies are 
being continually refined to promote the efficiency of 
the program. Future evaluation will be required to 
determine the efficacy of changes instituted in the past. 



186 



APPENDIX I 
PSRO Implementation Status 



187 



Appendix I, PSRO Implementation Status 

Column Column Title 

(1) Area funding status (P = planning, 

C ^ conditional, 

U = unfunded) 12/30/79. 

(2) Date of conditional contract award. 

(3) Percent of hospitals implemented, 6/30/78. 

(4) Percent of implemented hospitals under fully 
delegated review, 6/30/78. 

(5) Percent of implemented hospitals under 
partially delegated review, 6/30/78. 

(6) Percent of implemented hospitals under 
non-delegated review, 6/30/78. 

(7) Percent of hospitals implemented, 6/30/79. 

(8) Percent of implemented hospitals under 
fully delegated review, 6/30/79. 

(9) Percent of implemented hospitals under 
partially delegated review, 6/30/79. 

(10) Percent of implemented hospitals under 
non-delegated review, 6/30/79. 

(11) Percent of reviewed Federal discharges under 
fully delegated review, 6/30/79. 

(12) Percent of reviewed Federal discharges 
under partially delegated review, 6/30/79. 



(13) 

(14) 

(15) 
(16) 
(17) 
(18) 



Percent of reviewed Federal discharges under 

non-delegated review. 

Percent of eligible physicians who are 

members of the PSRO, 1978. 

Long-term care review demonstration site. 

Ambulatory care review demonstration site. 

Ancillary review special initiative site. 

Physician services review demonstration site. 



Sources 

Columns 1-2 = HSQB Division of Program 

Operation 
Columns 3-10 = HSQB Implementation Status 

Report 
Columns 11-13 = Form 121 Summaries, HSQB 

Division of Data Planning 
Column 14 := PSRO grant applications and 

contract proposals 
Columns 15-18 = HSQB Division of Program 

Operations. 



Note 

NA- 



-Not Available 



Appendix I, PSRO Implementation Status 



PRSO 



8 



10 



11 12 13 14 15 16 17 18 



4AL01 


C 


6/1/76 


100 


68 





32 


98 


68 





32 


91 





9 


56 


X 


0AK01 


C 


6/30/76 


71 


76 





24 


96 


83 





17 


96 





4 


65 




9AZ01 


C 


4/1/79 






















NA 


NA 


NA 


9 




9AZ02 


C 


9/30/77 


NA 


NA 


NA 


NA 


60 


86 


7 


7 


92 





8 


45 




6AR01 


C 


6/27/75 


90 


99 





1 


100 


95 





5 


98 





2 


63 




9CA01 


C 


6/26/75 


100 


38 


58 


4 


100 


40 


44 


16 


51 


39 


10 


61 


X 


9CA02 


C 


3/10/78 


NA 


NA 


NA 


NA 


100 


37 





59 


57 


3 


40 


41 




9CA03 


C 


6/26/75 


100 


42 


42 


16 


100 


39 


46 


15 


29 


71 





63 




9CA04 


C 


4/1/75 


100 


9 





91 


100 


13 





87 


NA 


NA 


NA 


72 


X 


9CA05 


C 


9/1/76 


95 


100 








100 


100 








100 








59 


X 


9CA06 


C 


9/30/78 


NA 


NA 


NA 


NA 


60 


100 








100 








42 




9CA07 


C 


7/1/79 






















NA 


NA 


NA 


29 




9CA08 


C 


6/29/74 


100 





57 


43 


100 





50 


50 





78 


22 


70 


X 


9CA09 


C 


6/29/76 


100 


85 





15 


100 


85 





15 


83 





17 


60 




9CA10 


C 


9/30/77 


100 


65 


29 


6 


100 


65 


29 


6 


72 


26 


2 


56 




9CA11 


C 


1/1/79 






















NA 


NA 


NA 


47 




9CA12 


C 


6/26/75 


100 


36 


46 


18 


100 


36 


46 


18 


46 


51 


3 


68 




9CA13 


C 


6/30/78 


NA 


NA 


NA 


NA 


100 


77 





23 


97 





3 


91 




9CA14 


C 


5/27/76 


100 


62 


7 


31 


100 


62 


7 


31 


90 


1 


9 


72 




9CA15 


C 


3/30/78 


NA 


NA 


NA 


NA 


73 


100 








100 








45 




9CA16 


C 


3/29/77 


93 


92 


8 





100 


79 


7 


14 


97 


2 


1 


70 




9CA17 


C 


6/30/76 


100 


82 





18 


100 


75 





25 


93 





7 


86 




9CA18 


C 


12/30/77 


44 


17 





63 


100 


44 





56 


46 


9 


45 


33 




9CA19 


U 










s 










NA 


NA 


NA 


25 




9CA20 


C 


9/21/76 


89 


100 








97 


94 


3 


3 


97 





3 


56 




9CA21 


c 


9/30/77 


21 


83 





17 


93 


38 


8 


54 


36 


9 


55 


26 




9CA22 


c 


6/23/76 


83 


100 








100 


71 





29 


93 





7 


79 




9CA23 


c 


1/5/77 


100 


42 





58 


76 


28 


68 


4 


NA 


NA 


NA 


48 




9CA24 


c 


6/30/75 


93 


100 








100 


82 





18 


95 


5 





42 




9CA25 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


33 





67 


89 





11 


32 




9CA26 


c 


4/1/79 


' 




















NA 


NA 


NA 


26 





188 



X X 



X 
X 



X 
X 
X 



J 



Appendix I, PSRO Implementation Status (cont'd) 



PRSO 


1 


2 


3 


4 


5 


6 


7 


8 


9 


10 


11 


12 


13 


14 


15 16 17 18 


9CA27 


C 


6/26/75 


100 


87 





13 


100 


87 





13 


94 





6 


83 






9CA28 


C 


4/1/79 










21 


88 





12 


100 








41 






8CO01 


C 


6/28/74 


100 


5 





95 


97 


6 





94 


16 





84 


54 


X 




1CT01 


C 


6/30/75 


100 


100 








100 


100 








100 








93 


X 




1CT02 


C 


6/28/75 


100 


92 





8 


100 


92 





8 








100 


63 


X 




1CT03 


C 


6/27/75 


100 


89 





11 


100 


89 





11 


99 





1 


78 


X 




1CT04 


C 


6/30/75 


100 


100 








100 


100 








100 








75 


X 




3DE01 


C 


6/1/76 


100 


72 


14 


14 


100 


72 


14 


14 


48 


52 





NA 


X 


X 


3DC01 


C 


6/28/75 


100 


20 


33 


47 


82 


14 


50 


36 


12 


57 


31 


47 


X : 


K X 


4FL01 


C 


9/30/78 


NA 


NA 


NA 


NA 














NA 


NA 


NA 


64 






4FL02 


C 


3/30/78 


NA 


NA 


NA 


NA 


94 


53 





47 


84 





16 


55 




X 


4FL03 


C 


3/30/77 


100 


100 








100 


100 








100 








82 


X 




4FL04 


P 




















NA 


NA 


NA 


NA 






4FL05 


C 


6/30/78 


NA 


NA 


NA 


NA 


65 


100 








100 








46 






4FL06 


C 


1/1/79 










46 


100 








83 





17 


60 






4FL07 


C 


1/1/79 










93 


79 





21 


68 





32 


31 




X 


4FL08 


P 




















NA 


NA 


NA 


NA 






4FL09 


C 


1/1/79 










80 


100 








54 





46 


42 






4FL10 


C 


9/30/78 


NA 


NA 


NA 


NA 


65 


100 








100 








47 






4FL11 


c 


6/30/78 


NA 


NA 


NA 


NA 


19 


100 








80 





20 


34 






4FL12 


c 


12/15/75 


100 


82 


5 


13 


100 


73 


10 


17 


74 


14 


12 


35 


X X 


4GA01 


c 


7/1/79 






















NA 


NA 


NA 


34 






9Hi01 


c 


6/23/76 


47 


88 


12 





92 


79 


13 


8 


100 








57 






0ID01 


c 


6/26/75 


100 


8 





92 


100 


8 





92 


29 





71 


41 






5IL01 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


53 





47 


83 





17 


54 




X 


5IL02 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


85 





15 


77 





23 


36 




X 


5IL03 


c 


9/30/76 


100 


93 





7 


100 


92 





8 


85 





15 


41 






5IL04 


c 


6/26/75 


100 


83 





17 


100 


83 





17 


NA 


NA 


NA 


60 


X 




5IL05 


c 


3/30/78 


72 


83 





17 


100 


88 





12 


89 





11 


43 


X 


X 


5IL06 


c 


6/30/78 


NA 


NA 


NA 


NA 


100 


83 





17 


95 





5 


50 






5IL07 


c 


3/1/79 










18 


100 








100 








48 






5IL08 


c 


3/30/78 


57 


100 








83 


82 





18 


75 


16 


9 


58 






5IN01 


u 


9/30/77 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


51 






5IN02 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


62 





38 


61 





39 


32 




X 


5IN03 


c 


6/30/78 


NA 


NA 


NA 


NA 


92 


50 





50 


NA 


NA 


NA 


41 






5IN04 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


31 





69 


84 





66 


24 






5IN05 


c 


9/30/77 


NA 


NA 


NA 


NA 


100 


82 





18 


85 





15 


39 






5IN06 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


80 





20 


70 





30 


44 






5IN07 


c 


6/30/78 


NA 


NA 


NA 


NA 


100 


100 








100 








75 




X 


7IA01 


c 


12/15/75 


63 


100 








100 


93 





7 


96 


1 


3 


61 


X 


X 


7KS01 


c 


9/30/77 


13 


100 
















100 








52 






4KY01 


c 


9/15/76 


98 


71 





29 


100 


71 





29 


70 





30 


59 


X 


X 


6LA01 


c 


9/30/78 


NA 


NA 


NA 


NA 


48 


4 





96 


3 





97 


20 






6LA02 


c 


9/30/78 


NA 


NA 


NA 


NA 


40 


36 





64 


40 





60 


16 






6LA03 


c 


9/30/78 


NA 


NA 


NA 


NA 


35 








100 








100 


16 






6LA04 


c 


9/30/78 


NA 


NA 


NA 


NA 


34 


8 





92 


11 





89 


27 






1ME01 


c 


6/27/75 


93 


94 





6 


96 


92 





8 


97 





3 


55 






3MD01 


c 


6/30/76 


100 


100 








100 


100 








100 








61 


X 




3MD02 


c 


6/27/75 


100 


94 





6 


100 


94 





6 


94 





6 


NA 


X 


X 


3MD03 


c 


6/27/75 


100 


50 





50 


100 


50 





50 


NA 


NA 


NA 


55 


X 


X 


3MD04 


c 


6/29/74 


100 


17 





83 


100 


17 





83 


21 





79 


52 


X 


X 


3MD05 


c 


6/27/75 


100 


100 








100 


100 








100 








75 


X 




3MD06 


c 


6/27/75 


100 


40 


40 


20 


100 


40 


40 


20 


72 


15 


13 


70 






3MD07 


c 


6/27/75 


100 


17 


83 





100 


83 





17 


NA 


NA 


NA 


68 


X 


X 


1MA01 


c 


6/27/75 


100 


100 








100 


94 





6 


98 





2 


51 


X 


X 


1MA02 


c 


6/1/76 


100 


94 





6 


100 


94 





6 


97 





3 


82 


X 




1MA03 


c 


6/29/74 


100 


88 


12 





100 


100 








100 








85 


X 


X 



189 



Appendix I, PSRO Implementation Status (cont'd) 



PRSO 


1 


2 


3 


4 


5 


6 


7 


8 


9 


10 


11 


12 


13 


14 


15 16 17 18 


1MA04 


C 


6/28/74 


97 


66 


7 


27 


99 


65 


24 


11 


66 


23 


11 


60 






1MA05 


C 


6/30/75 


100 


93 


7 





100 


93 


7 





98 





2 


64 


X 




5MI01 


C 


6/30/75 


100 


5 





95 


95 


5 





95 


10 





90 


73 




X 


5MI02 


C 


7/1/79 


















93 





7 


48 






5MI03 


C 


3/30/78 


NA 


NA 


NA 


NA 


100 


86 





14 








50 




X 


5MI04 


C 


6/30/78 


NA 


NA 


NA 


NA 


92 


100 








100 








68 






5MI05 


C 


6/27/75 


100 


22 





78 


100 


11 





89 


4 





96 


60 




X 


5MI06 


C 


6/30/78 


NA 


NA 


NA 


NA 


100 


83 





17 


95 





5 


58 




X 


5MI07 


C 


3/30/78 


NA 


NA 


NA 


NA 


100 


87 





13 


89 





11 


49 




X 


5MI08 


C 


9/30/77 


NA 


NA 


NA 


NA 


71 


91 


9 





88 


7 


5 


79 






5MI09 


C 


9/30/78 


NA 


NA 


NA 


NA 


40 


100 








100 








NA 






5MI10 


C 


9/30/78 


NA 


NA 


NA 


NA 


42 


95 





5 


100 








54 






5MN01 


C 


6/28/74 


100 


100 








52 


100 








100 








68 


X 


X 


5MN02 


C 


6/30/76 


26 


100 








100 


94 


1 


5 


96 





4 


54 






4MS01 


C 


6/29/74 


100 


46 


46 


8 


100 


46 


47 


7 


NA 


NA 


NA 


NA 


X 




7MO01 


C 


9/30/76 


81 


100 








91 


86 


14 





91 


6 


3 


49 






7MO02 


C 


6/1/76 


100 


85 





15 


100 


84 





16 


79 





21 


54 




X 


7MO03 


C 


6/27/75 


83 


100 








97 


91 


6 


3 


96 





4 


42 


X 


X 


7MO04 


c 


7/22/76 


96 


91 





9 


100 


88 





12 


95 





5 


59 






7MO05 


c 


6/2/76 


100 


86 


10 


4 


100 


81 


14 


5 


94 


5 


1 


70 




X 


8MT01 


c 


5/8/75 


99 


5 


2 


92 


100 


5 


1 


94 


20 


2 


78 


44 


X 


X 


7NE01 


u 
































9NV01 


c 


6/30/76 


100 


78 


13 


9 


100 


100 








99 


1 





53 






1NH01 


c 


6/27/75 


100 


89 





11 


93 


92 





8 


99 





1 


56 


X 


X 


2NJ01 


c 


5/28/76 


100 


100 








100 


100 








100 








46 




X 


2NJ02 


c 


6/27/75 


100 


14 


86 





100 


14 


86 





17 


83 





54 




X 


2NJ03 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


91 





9 


98 





2 


31 






2NJ04 


c 


7/21/76 


82 


93 


7 





100 


93 





7 


96 


2 


2 


70 




X 


2NJ05 


c 


3/30/78 


NA 


NA 


NA 


NA 


82 


89 





11 


78 





22 


91 




X 


5NJ06 


c 


3/30/78 


NA 


NA 


NA 


NA 


70 


100 








100 








50 




X 


5NJ07 


c 


6/30/78 


NA 


NA 


NA 


NA 


68 


67 





33 


53 





47 


33 






5NJ08 


c 


9/21/76 


33 


100 








100 


95 





5 


95 





5 


70 




X 


5NM01 


c 


11/1/74 


100 


100 








98 


55 





45 


77 





23 


57 


X X 


2NY01 


u 


6/27/75 


84 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


62 






2NY02 


c 


6/27/75 


100 


NA 


NA 


NA 


100 


100 








100 








63 




X 


2NY03 


c 


9/21/76 


41 


NA 


NA 


NA 


100 


85 


11 


4 


81 


10 


9 


60 






2NY04 


c 


6/30/76 


46 


100 








100 


77 


23 





82 





18 


83 


X 


X 


2NY05 


c 


6/28/75 


87 


100 








93 


93 


7 





100 








75 






2NY06 


c 


3/30/78 


NA 


NA 


NA 


NA 


74 


100 








NA 


NA 


NA 


75 


X 




2NY07 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


88 


6 


6 


99 





1 


43 






2NY08 


c 


3/30/78 


NA 


NA 


NA 


NA 


95 


90 


5 


5 


89 


11 





47 






2NY09 


c 


6/27/75 


100 


100 








100 


100 








NA 


NA 


NA 


57 




X 


2NY10 


c 


6/27/75 


100 


100 








100 


100 








100 








76 




X 


2NY11 


c 


6/27/75 


83 


60 





40 


100 


59 


3 


38 


69 


12 


19 


47 




< 


2NY12 


c 


9/30/76 


100 


100 








100 


100 








100 








81 






2NY13 


c 


6/27/75 


96 


100 








100 


100 








100 








47 






2NY14 


c 


1/31/77 


86 


83 


6 


11 


100 


84 


11 


5 


93 


7 





54 


X 




2NY15 


c 


6/27/75 


10O 


100 








100 


100 








100 








36 




X 


2NY16 


c 


6/27/75 


100 


85 





15 


100 


67 





32 


70 





30 


70 


X : 


< 


2NY17 


c 


6/30/78 


NA 


NA 


NA 


NA 


42 


100 








81 





19 


40 






4NC01 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


75 


3 


22 


84 


12 


4 


62 




X 


4NC02 


c 


3/1/77 


91 


25 





75 


95 


11 





89 


25 





75 


68 


X 


X 


4NC03 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 








100 


81 





19 


52 




X 


4NC04 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


72 


14 


14 


97 





3 


50 




X 


4NC05 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


43 





57 


43 





57 


37 




X 


4NC06 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


50 


1 


50 


58 





42 


53 




X 


4NC07 


c 


9/30/77 


NA 


NA 


NA 


NA 


18 


20 


40 


40 


100 








NA 




X 



190 



Appendix I, PSRO Implementation Status (cont'd) 



PRSO 


1 


2 


3 


4 


5 


6 


7 


8 


9 


10 


11 


12 


13 


14 


15 16 17 18 


4NC08 


C 


3/30/78 


NA 


NA 


NA 


NA 


100 


64 





36 


70 





30 


NA 




X 


8ND01 


c 


9/22/76 


100 


31 


60 


9 


100 


33 


58 


9 


69 


29 


2 


75 




X 


5OH01 


c 


6/27/75 


100 


92 





8 


100 


92 





8 


96 





4 


59 






5OH02 


c 


9/9/76 


100 


93 





7 


100 


93 





7 


98 





2 


54 


X 




5OH03 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


59 


23 


18 


77 


16 


7 


69 


X 


X 


5OH04 


c 


6/26/75 


100 


84 





16 


90 


63 





37 


64 





36 


71 


X 




5OH05 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


50 





50 


50 





50 


46 






5OH06 


c 


9/30/77 


100 








71 


64 


100 








100 








45 


X 




5OH07 


c 


6/30/78 


NA 


NA 


NA 


NA 


100 


78 





22 


77 





23 


37 






5OH08 


c 


6/30/78 


NA 


NA 


NA 


NA 


100 


100 








100 








78 


X 




5OH09 


c 


3/30/78 


NA 


NA 


NA 


NA 


100 


100 








100 








79 






5OH10 


c 


9/1/76 


100 


76 





24 


100 








100 


89 





11 


52 


X 


X 


50H11 


c 


3/30/78 


NA 


NA 


NA 


NA 


85 


100 








100 








50 






50H12 


c 


6/30/76 


100 


77 





23 


100 


76 





24 


78 





22 


37 






50H13 


c 


7/1/79 






















NA 


NA 


NA 


16 






6OK01 


c 


3/30/78 


91 


100 








91 


100 








100 








37 






0OR01 


c 


6/28/74 


100 


54 


31 


15 


100 


54 


31 


15 


NA 


NA 


NA 


64 


X 




0OR02 


c 


6/30/75 


89 


98 





2 


100 


98 


2 





100 








65 


X 




3PA01 


c 


5/8/75 


NA 


NA 


NA 


NA 


100 


100 








NA 


NA 


NA 


67 






3PA02 


c 


6/3/76 


79 


100 








95 


100 








99 


1 





68 






3PA03 


c 


3/30/78 


NA 


NA 


NA 


NA 


25 


100 








NA 


NA 


NA 


32 






3PA04 


c 


6/1/76 


91 


90 


10 





100 


82 





18 


83 


17 





76 






3PA05 


c 


9/30/77 


50 


100 








100 


94 





6 


100 








53 






3PA06 


c 


6/27/75 


100 


93 





7 


100 


93 





7 


94 





6 


NA 






3PA07 


c 


6/27/75 


100 


83 





17 


100 


82 





18 


83 





17 


49 




X 


3PA08 


c 


5/30/76 


100 


100 








100 


100 








100 








48 




X 


3PA09 


c 


6/27/75 


100 


100 








100 


100 








100 








57 






3PA10 


c 


9/29/78 


NA 


NA 


NA 


NA 


15 


100 








NA 


NA 


NA 


40 






3PA11 


c 


6/27/75 


71 


100 








85 


100 








100 








61 






3PA12 


c 


6/27/75 


100 


84 





16 


100 


84 





16 


92 





8 


54 






2PR01 


c 


9/30/77 


4 


100 








81 


60 


15 


25 


NA 


NA 


NA 


48 






1RI01 


c 


6/30/75 


100 


100 








100 


100 








100 








NA 


X 




4SC01 


c 


6/27/75 


100 


27 





73 


100 


41 





59 


64 





36 


69 


X 


X 


8SD01 


c 


6/4/76 


100 


52 





48 


100 


52 





48 


78 





22 


70 






4TN01 


c 


4/1/77 


56 


100 








96 


90 


10 





NA 


NA 


NA 


5 






4TN02 


u 


6/29/74 


100 


35 


31 


34 


NA 


NA 


NA 


NA 


NA 


NA 


NA 


NA 






6TX01 


P 




















NA 


NA 


NA 


18 






8UT01 


c 


6/18/74 


100 








100 


100 








100 








100 


69 


X ; 


< 


1VT01 


c 


6/1/76 


78 


100 








100 


100 








100 








74 


X 




2VI01 


c 


6/30/78 


NA 


NA 


NA 


NA 


100 








100 








100 


63 






3VA01 


c 


9/30/77 


44 


100 








100 


88 





12 


88 





12 


55 


X 


X 


3VA02 


c 


1/3/77 


100 


100 








100 


100 








100 








48 


X 




3VA03 


c 


6/30/78 


NA 


NA 


NA 


NA 


29 


100 








100 








53 






3VA04 


c 


6/30/78 


NA 


NA 


NA 


NA 


83 


100 








100 








39 






3VA05 


c 


1/5/77 


95 


95 





5 


97 


64 


24 


12 


91 





7 


57 


X 




0WA01 


c 


6/30/75 


100 


95 





5 


100 


95 





5 


94 





6 


58 




X X 


3WV01 


c 


6/28/75 


100 


71 





29 


100 


71 





29 


91 





9 


57 






5WI01 


c 


6/30/75 


100 


95 





5 


63 


89 





11 


95 





5 


61 






5WI02 


c 


6/30/75 


100 


100 








91 


100 








100 








55 






8WI01 


c 


6/28/74 


100 








100 


100 








100 





100 





93 


X 





191 



APPENDIX II 

Glossary of Abbreviations and 
Acronyms Used in this Volume 



I 



192 



Glossary of Abbreviations and Acronyms 
Used in this Volume 

AAPSRO American Association of Professional 

Standards Review Organizations 
AB Aid to the Blind 

AC Admission Certification 

ACR Ambulatory Care Review 

AD Aid to the Disabled 

AEP Appropriateness Evaluation Protocol 

AFDC Aid to Families with Dependent Children 

AlP Annual Implementation Plan 

ALOS Average Length of Stay 

AMA American Medical Association 

APTD Aid to the Permanently and Totally 

Disabled 
ARF Area Resource File 

ASR Ancillary Services Review 

AUTOGRP Autogroup (See Glossary of Terms, 

Appendix III) 
BC/BS Blue Cross/Blue Shield 

BHI Bureau of Health Insurance (Social 

Security Administration) 
BQA Bureau of Quality Assurance 

CLTR Care Level Timeliness Review 

CME Continuing Medical Education 

CPHA Commission on Professional and Hospital 

Activities 
OPT Current Procedural Terminology 

CR Concurrent Review 

CSR Continued Stay Review 

DDR Division of Direct Reimbursement 

(Social Security Administration) 
DHEW Department of Health, Education and 

Welfare 
DISC Discharges (per 1000 enrollees) 

DO Doctor of Osteopathy 

DOC Days of Care (per 1,000 enrollees) 

DRG Diagnostically Related Group 

DS Discharge Screening 

ECF Extended Care Facility 

ELOS Estimated Length of Stay 

EMCRO Experimental Medical Care Review 

Organization 
ESRD End Stage Renal Disease 

Fl Fiscal Intermediary 

FMC Foundation for Medical Care 

FRM Federal Reports Manual 

FY Fiscal Year 

GAO General Accounting Office 

HCFA Health Care Financing Administration 

HDS Hospital Discharge Survey (National 

Center for Health Statistics) 
H-ICDA Hospital-International Classification of 

Diseases-Adapted 
HMO Health Maintenance Organization 

HSA Health Systems Agency 

HSP Health Systems Plan 

HSQB Health Standards and Quality Bureau 

HUP Hospital Utilization Project (Western 

Pennsylvania) 



ICDA-8 

ICF 
ICF-MR 

lOM 

IS 

ISD 

JCAH 

LOS 

LTC 

MAP 

MCE 

MCH 

MD 

MEDPAR 

MF! 

MMIS 

MOU 
MRB 
NCHS 
NCHSR&D 

NPSRC 

OAA 
CASH 

OB 

0MB 

OPEL 

OPPR 
OPSRO 

OQS 
ORDS 

PAS 

PHDDS 

PHS 

PIP 

PIPSRO 

PL 

PMIS 

PSRO 

RFP 
SI 

SNF 
SPSRC 

SRS 

SSA 

SSI 

TAP 

UHDDS 

UR 



International Classification of Diseases- 
Adapted (8th Revision) 
Intermediate Care Facility 
Intermediate Care Facility for the 

Mentally Retarded 
Institute of Medicine (National Academy 

of Sciences) 
Intensity of Services 
Intensity, Severity, and Discharge 
Joint Commission on the Accreditation 

of Hospitals 
Length of Stay 
Long Term Care 
Medical Audit Program 
Medical Care Evaluation (Studies) 
Maternal and Child Health (Program) 
Doctor of Medicine 

Medicare Provider Analysis and Review 
Master Facility Inventory (National 

Center for Health Statistics) 
Medicaid Management Information 

System 
Memorandum of Understanding 
Medical Review Board 
National Center for Health Statistics 
National Center for Health Services 

Research and Development 
National Professional Standards Review 

Council 
Old Age Assistance 
Office of the Assistant Secretary for 

Health 
Obstetric 

Office of Management and the Budget 
Office of Planning, Evaluation, and 

Legislation 
Office of Policy, Planning, and Research 
Office of Professional Standards Review 

Organizations 
Office of Quality and Standards 
Office of Research, Demonstrations, and 

Statistics 
Professional Activity Study 
PSRO Hospital Discharge Data Set 
Public Health Service 
Periodic Interim Payment 
Private Initiative in PSRO 
Public Law 

PSRO Management Information System 
Professional Standards Review 

Organization 
Request for Proposal 
Severity of Illness 
Skilled Nursing Facility 
Statewide Professional Standards 

Review Council 
Social and Rehabilitative Services 
Social Security Administration 
Supplemental Security Income 
Technical Advisory Panel 
Uniform Hospital Discharge Data Set 
Utilization Review 



193 



State Abbreviations 



Alabama AL 

Alaska AK 

Arizona AZ 

Arkansas AR 

California CA 

Colorado CO 

Connecticut CT 

Delaware DE 

Dist. of Columbia DC 

Florida FL 

Georgia GA 

Hawaii HI 

Idaho ID 

Illinois IL 

Indiana IN 

Iowa lA 

Kansas KS 

Kentucky KY 

Louisiana LA 

Maine ME 

Maryland MD 

Massachusetts MA 

Michigan Ml 

Minnesota MN 

Mississippi MS 

Missouri MO 

Montana MT 



Nebraska NE 

Nevada NV 

New Hampshire NH 

New Jersey NJ 

New Mexico NM 

New York NY 

North Carolina NC 

North Dakota ND 

Ohio OH 

Oklahoma OK 

Oregon OR 

Pennsylvania PA 

Puerto Rico PR 

Rhode Island Rl 

South Carolina SC 

South Dakota SD 

Tennessee TN 

Texas TX 

Utah UT 

Vermont VT 

Virgin Islands VI 

Virginia VA 

Washington WA 

West Virginia WV 

Wisconsin Wl 

Wyoming WY 



194 



APPENDIX III 
Glossary of Terms 



195 



Admission Certification (AC) — One segment of the 
Concurrent Review component of the model PSRO 
review system in which Federal patients' admissions 
are reviewed for medical necessity and the 
appropriateness of the level of care, using diagnosis- 
specific admission criteria and standards, plus the 
assignment of a certified length of stay using 
diagnosis-specific norms, criteria, and standards 

Admission Rate — The number of admissions to a 
health care facility or group of facilities divided by 
the total relevant population. 

Ambulatory Care — Health care services not provided 
on an inpatient basis. 

Ancillary Services — Those health care services provided 
in a health care facility other than general nursing 
care, such as laboratory, radiology, physical therapy, 
respiratory therapy, etc. 

Annual Implementation Plan (AIR) — a plan which the 
National Health Planning and Resources Development 
Act of 1974 (P. L. 93-641) requires health systems 
agencies to prepare or undate annually. Specifying, 
describing how to implement, and giving priority to 
short-run objectives which will achieve the long range 
goals of the agency, detailed in its health system plan. 

Area Resource File (ARF) — A county-based file of area 
population, demographic, and health services data. 

Assessment and Monitoring — Those procedures 
performed by a fiscal intermediary or State agency 
to evaluate the review procedures of a PSRO or 
hospital and to verify the results of a sample of the 
reviews. 

Assignment — An agreement in which a patient assigns 
to another party, usually a provider, the right to 
receive payment from a third-party payor. It is used 
instead of a patient paying directly for the service 
rendered and then receiving reimbursement from 
public or private insurance program. 

Autogroup (AUTOGRP) — An interactive statistical 
computer system for the analysis of health care 
data, specifically used to develop diagnostically 
related groups (DRGs) for comparative analyses. 
AUTOGRP was developed by the Center for the Study 
of Health Services at Yale University with Federal 
support. 

Average Length of Stay (ALOS) — A measure of the 
number of days stay averaged over all patients; 
calculated by dividing the total days of care by the 
number of discharges. 

Baseline Data — Data representing the pre-treatment 
period(s) of a study. 

Benefit-cost — The ratio of the dollar value of the impact 
of a program or activity to the costs incurred in 
carrying out the program or activity. 

Break-even — The benefit or return from a program or 
activity equals the costs incurred. 

Break-even point — The combination of inputs which 
brings about an equal return (output). 

Case mix — The relative distribution of discharges of 
different diagnoses or diagnostic groups. 

Case mix severity — A measure of the relative distribution 
of discharges of different diagnoses measured by 
the aggregate expected days of care (calculated from 



expected lengths of stay attributed to the diagnoses 
of actual cases). 

Claims Review — The review of claims for reimbursement 
by a third party payor, generally for eligibility and 
coverage determinations, but sometimes including 
medical necessity. 

Certified Days — Hospital inpatient days of care that are 
certified by a PSRO for payment at the hospital 
level. 

Concurrent Review (CR) — One component of the model 
PSRO review system where institutionalization of 
federal patients is reviewed while the patient is in 
the facility. CR is composed of Admission Certification 
and Continued Stay Review. 

Conditional PSRO — An organization that having 
demonstrated its capability to perform the duties 
of fully designated PSRO, enlisted the membership of 
at least 25% of the physicians in the area, 
developed a review and implementation plan, and 
met other PSRO organizational requirements, has 
been awarded a contract to implement the required 
review and other functions in its designated area. 

Continued Stay Review (CSR) — One segment of the 
Concurrent Review component of the model PSRO 
review system in which Federal patients whose 
admission has been certified by the Admission 
Certification process are periodically reviewed to 
assure the need for continued hospital care and that 
care is being provided efficiently and economically 
using diagnosis-specific norms, standards, and 
criteria. 

Criteria — Predetermined elements of care against which 
aspects of the quality of a medical service may be 
compared; developed by professionals relying on 
professional expertise and on the professional 
literature. 

Current Procedural Terminology (CPT) — A system of 
terminology and coding developed by the American 
Medical Association that is used for describing, 
coding and reporting medical services and procedures. 

Days of Care (DOC) — A measure of utilization equal 
to the sum of the number of days stay for all 
inpatients; also equal to the product of the number 
of admissions and the average length of stay. The 
day of admission is counted as a day of care; the 
day of discharge is not. 

Delegation — The assignment of Concurrent Review, 
Medical Care Evaluation Study, and/or Reconsideration 
functions to a hospital committee of physicians that 
has demonstrated a capability and desire to perform 
those functions. 

Discharge abstract service — A data service in which 
participating provider institutions submit discharge 
abstracts which are then processed and reformatted 
to provide various utilization profiles. 

Elective Surgery — Surgery Which need not be performed 
on an emergency basis, because reasonable delays 
will not affect the outcome of surgery. 

Federal patient — A patient whose care is paid for in 
whole or in part under the Social Security Act 
(Title V, XVIII, or XIX). 



196 



Fixed costs — That portion of total costs which does not 
vat7 with changes in quantity. 

Focused review — f?eview that concentrates on a 
perceived problem area that may be a specific 
diagnosis, procedure, provider, or other limited scope 
topic, in lieu of comprehensive review. 

Fully Designated PSRO — A PSRO that has performed 
satisfactorily as a conditional PSRO, and has been 
granted full PSRO status as specified in the law. 

Health Maintenance Organization (HMO) — A prepaid 
group practice plan. 

Health System Agency (HSA) — A regionally-based 
planning organization established pursuant to P.L. 
93-641 and charged to plan for the health needs 
of all residents of its health service area, promote 
the development of needed health resources, and 
reduce inefficiencies. 

Health Systems Plan (HSP) — A long range health plan 
prepared by a health systems agency for its health 
service area specifying the health goals considered 
appropriate by the agency for the area. 

Intensity of Service (IS) — The quantities of services 
provided to patients in a hospital or some identifiable 
setting. 

Joint Commission on Accreditation of Hospitals 
(JCAH) — A private nonprofit organization whose 
purpose is to encourage the attainment of uniformly 
high standards of institutional medical care 
through surveys, accreditation programs, and the 
establishment of guidelines for the operation of 
hospitals and other health facilities. 

Length of Stay (LOS) — A measure of utilization equal 
to the number of days a patient is institutionalized; 
calculated by subtracting the admission date from 
the discharge date. 

Master Facility Inventory (MFI) — A data file prepared by 
the National Center for Health Statistics of 
information on individual hospitals and nursing 
homes. 

Medicaid State Agency — A State governmental agency 
that administers the State's Medicaid program and 
normally acts as an intermediary for the payment 
of Medicaid claims. 

Medical Care Evaluation (MCE) Studies — One 

component of the model PSRO review system whereby 
a medical care topic area is selected and studied 
on a sample basis, usually retrospectively, to assure 
the quality of care and effective administration of 
health care services. Generally the process involves 
choosing a topic, establishing criteria and standards, 
collecting and analyzing data to identify deficiencies, 
implementing corrective action, and restudying the 
topic. 

Medicare Carrier — A commercial health insurer or an 
organization with which the Supplemental Medical 
Insurance (Part B) of the Medicare program contracts 
for administration of various functions, including 
payment of cliams. 

Medicare fiscal intermediary — A public or private agency 
or organization selected by providers of health care 
which enters into an agreement with the Secretary 
of HEW under Hospital Insurance Program (Part A) 



of Medicare, to pay claims and perform other 
functions for the Secretary with respect to such 
providers. 

Medicare Part A — Part of Title XVIII of the Social 
Security Act providing hospital insurance benefits to 
the aged and disabled covering hospitalization and 
related post-hospital services. 

Medicare Part B — Part of Title XVIII of the Social 
Security Act providing voluntary supplemental 
medical insurance benefits to the aged and disabled 
covering physician and other professional services, 
outpatient services, home health services, etc. 

Norms — Numerical or statistical measures of usual 
observed performance. 

Objective setting — Part of HSQB's program initiatives 
whereby PSROs set performance objectives to enhance 
impact and control costs. 

Occupancy rate — A measure of inpatient health facility 
use determined by dividing available bed days by 
patient days. It measures the average percentage 
of a hospital's beds occupied and may be institution- 
wide or specific for one department or service. 

Patient day — A unit of utilization consisting of one day 
of stay for one patient. 

Per Diem Charge — The average total charge for a day 
of care (one patient day), including both accommoda- 
tions and ancillary charges. 

Planning PSRO — An organization which has demon- 
strated the potential to meet the requirements for 
conditional PSRO designation and has been awarded 
a contract for financial assistance to develop that 
potential by developing a review plan and imple- 
mentation plan, recruiting physician members, and 
collecting baseline data. 

Pre-admission Certification — Performance of the 

Admission Certification process on elective cases prior 
to admission. Pre-admission certification may be 
performed by PSROs in lieu of Admission Certification 
with the approval of DHEW. 

Principal Procedure — The primary non-diagnostic 
surgical procedure listed on the billing form using 
Current Procedural Terminology (CRT) for coding. 

Primary Discharge Diagnosis — The primary reason for 
patient's hospitalization — that is, the disease, illness, 
condition or injury requiring the current hospitalization 
— recorded on the billing form (SSA-1453). 

Professional Activity Study — A shared-computer medical 
record information system purchased by hospitals 
from the Commission on Professional and Hospital 
Activities( CPHA), a nonprofit organization located 
in Ann Arbor, Michigan. 

Profile Analysis — One component of the model PSRO 
review system whereby aggregated statistical data are 
used to identify areas where utilization practices may 
be inappropriate, to focus Concurrent Review and 
to assist in the selection of topics for Medical Care 
Evaluation studies. Additional purposes are to monitor 
the effectiveness of hospital review activities and 
display local, regional, and national norms of 
utilization. 



197 



Prospective Reimbursement — Any method of paying 
hospitals or other health programs in whicii amounts 
or rates of payment are established in advance for 
the coming year and the programs are paid these 
amounts regardless of the costs they actually incur. 
These systems of reimbursement are designed to 
introduce a degree of constraint on charge or cost 
increases by setting limits on amounts paid during 
a future period. 

Provider of Service File — A file which contains 

information on every hospital which is approved for 
participation in the Medicare program. 

Query File — The collection of unit record admission 
notices which are sent to the Social Security 
Administration by a hospital upon the admission 
of a patient identified as being eligible for Medicare 
coverage. 



Retrospective Review — Review performed after the 

discharge of a patient. 
Standards — Professionally developed expressions of 

the range of acceptable variation from a norm or 

criterion. 
Targeted Review — Review of less than 100% of patient 

stays. Usually directed toward diagnoses subject to 

abuse or which otherwise require special focus. 
Title V — The Maternal and Child Health and Crippled 

Children's Services Program of the Social Security 

Act. 

Title XVIIi — The Medicare (Health Insurance for the 
Aged and Disabled) Program of the Social Security 
Act. 

Title XIX — The Medicaid (Grants to States for Medical 
Assistance) Program of the Social Security Act. 



•U.S. GOVERNMENT PfilNTIHG OFFICE : 1980 0-311-168/1*18 



198 



Ci-IS Library 

C2-07-13 

7500 Socurity Eivu. _ 

:-:,::iitirr!ore,. Marvi-a no 21244 



Health Care 

Financing 

Research 

Report 



U.S. Department of Health, 
Education, and Welfare 

Patricia Roberts l-larris, Secretary 

Health Care Financing Administration 

Leonard D. Scliaeffer, Administrator 

Office of Research, Demonstrations, and 
Statistics 

James M. Kaple, Acting Director 

Diane Bolay, Director, Program Planning 
and Support 

Research Publications Group 

Karen Pelham O'Steen, Publications Coordinator 
Donna L. Esl<ow, Writer-editor 
Alice L Young, Writer-editor 
Cynttiia Dingle, Editorial Assistant 

Send changes of address or requests for this publication to: 
ORDS Publications, Rnn 1E9 Oak Meadows Building, 6340 Securi- 
ty Blvd, Baltimore, MD 21235. 



ens LIBRARY 



DEPARTMENT OF 

HEALTH AND HUMAN SERVICES 

HEALTH CARE FINANCING ADMINISTRATION 

BALTIMORE, MARYLAND 21235 




3 amS DDDD7537 D 

POSTAGE AND FEES PAID 

U.S. DEPARTMENT OF H.H.S. 

HHS-397 



OFFICIAL BUSINESS 

PENALTY FOR PRIVATE USE, S300